A method and system for optimizing an attraction tolling strategy under traffic demand management

Abstract

This invention provides an optimization method and system for attraction-based tolling strategies under traffic demand management. The method includes establishing a two-level programming model. The upper-level model aims to maximize the profit of the attraction points, and the decision variable of the upper-level model is the attraction point toll. By changing the attraction point toll, the generalized travel cost of travelers in the lower-level model is altered, thereby influencing travelers' route choices. The lower-level model is a traveler route selection model considering both attraction points and travel costs under stochastic demand. The system includes an information acquisition module, a model building module, and a model solving module. This invention transfers congestion tolling functions to the attraction points. Attraction points in congested areas charge positive tolls to travelers arriving at them, while attraction points in areas with excess road supply charge negative tolls. Travelers choose their routes based on the attraction point's attractiveness, the attraction point toll, and road segment impedance, thereby alleviating road segment congestion and making the road network traffic distribution more balanced.

Description

Technical Field

[0001] This invention relates to the field of transportation, and in particular to an optimization method and system for attraction-based pricing strategies under traffic demand management. Background Technology

[0002] In recent years, accelerated urbanization and the growth in the number of motor vehicles have exacerbated traffic congestion. Scholars have previously proposed measures to increase road supply, such as road expansion. However, Downs' theorem states that the growth rate of traffic demand often exceeds the growth rate of traffic supply, meaning that increased infrastructure construction leads to even greater traffic demand. Therefore, traffic demand management is increasingly being used as a strategy to alleviate congestion. Traffic demand management includes measures such as traffic restrictions, congestion pricing, and parking fees.

[0003] Studies have shown that traffic restrictions have a more significant short-term effect because high-income residents will buy two or more cars and drive the unrestricted car when one is restricted. In the long run, the effect of alleviating congestion is not obvious.

[0004] For congestion-based tolling, the most common implementation methods are setting up toll booths on congested sections for manual payment or installing electronic toll collection systems. Even though existing toll collection systems are relatively mature and the technical barriers to implementation have been greatly reduced, most methods still require the installation of a device in each vehicle capable of exchanging information with the metering system. Installing these devices in vehicles presents both a technical and economic challenge. Furthermore, coordination issues exist between different toll collection agencies, as drivers cannot equip themselves with different devices for every different toll road system they might use.

[0005] Compared to traffic demand management methods such as traffic restrictions and congestion pricing, parking fees are more widely accepted and less costly to implement. However, relying solely on parking fees offers a limited revenue stream; furthermore, many destinations offer free parking to attract customers, which can exacerbate congestion around those destinations.

[0006] Past research on parking fees has largely relied on statement preference surveys to study specific locations, such as workplaces, hospitals, tourist attractions, and street parking. Many cities typically price parking fees based on land scarcity, with limited literature considering factors like regional attractiveness and surrounding traffic congestion. Legorreta and Newmark's research found that only Nottingham, Perth, Sydney, Melbourne, and Singapore actually levy a zone tax on each parking space. Even when studies differentiate charging zones, they mostly divide them into suburbs and city centers. In reality, less attractive and less congested city centers don't need to charge high fees; conversely, more attractive and congested suburbs should charge fees to alleviate traffic problems. Arnott suggests that spatially differentiated parking fees can be as effective as temporally differentiated congestion pricing.

[0007] In fact, many places still offer free parking to travelers. In Cangzhou, China alone, nearly 100,000 public parking spaces are entirely free. A report by Brueckner and Franco states that over 80% of companies in the US provide parking for their employees, believing that free parking brings more traffic and thus greater revenue. While free parking seems to bring more revenue, in reality, the extra profits for operators are borne by society as a whole, creating significant negative externalities that extend far beyond the scope of a single operator. First, a large number of people driving increases road congestion, leading to negative externalities from congestion. Second, excessive parking also results in land loss, especially since the vast majority of car commuters drive alone. Not charging for parking may also raise fairness issues, as parkers are often wealthier than public transport users, yet they are unfairly sharing the costs of negative traffic externalities with all public transport users, even those who walk, take the bus, or cycle to the station (traffic congestion, air pollution, etc.). Calthrop points out that eliminating free parking at workplaces typically reduces the number of times people drive alone. Proost believes that shifting from employer-paid parking to employee-paid parking or workplace parking fees could encourage private car drivers to switch to public transportation. Since public transportation is more convenient in city centers, employees are more willing to choose to commute to work in the city center, which may even help enhance the city's agglomeration effect.

[0008] Therefore, it is also necessary to internalize the negative externalities of excess traffic demand by implementing parking fees and to differentiate parking fees by region. Summary of the Invention

[0009] To address the challenges of implementing congestion pricing, low public support, and the lack of regional differentiation in parking fees, this invention proposes an optimized method and system for attraction-based pricing strategies under traffic demand management. This method and system transfers congestion pricing functions to attraction areas, allowing them to set their own pricing (positive or negative fees, such as subsidies or discounts). For example, attraction areas in congested areas can increase parking fees or service charges; while those in areas with excess road supply can offer discounts on goods or lower entrance fees. Attraction areas, acting as rational economic agents, seek to maximize their profits. Travelers choose their routes based on attraction attractiveness, attraction pricing, and road segment resistance. Therefore, this method and system are more novel and more moderate.

[0010] The technical solution adopted in this invention:

[0011] This invention provides an optimization method for attraction-based pricing strategies under traffic demand management, the method comprising:

[0012] A two-level programming model is established, comprising an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of destinations (in a monopolistic market, all destinations in the city strive to maximize the common profit; in a monopolistic competition market (i.e., an oligopolistic market), each destination in the city strives to maximize its own profit). The decision variable of the upper-level model is the destination fee, which changes the generalized travel cost of travelers in the lower-level model, thereby influencing travelers' route choices. The lower-level model is a traveler route selection model considering both destinations and travel costs under stochastic demand.

[0013] Solving the bilevel programming model yields the optimal strategy for attraction pricing under traffic demand management;

[0014] The term "attraction" refers to facilities that are attractive to urban residents and tourists (attractions include attractions, shopping centers, leisure facilities, etc.);

[0015] The term "attraction charges" refers to the charging measures taken by an attraction to travelers arriving at that attraction. These include positive charges and negative charges (such as subsidies, discounts, etc.). Attractions in areas with congested roads charge positive charges to travelers arriving at them, while attractions in areas with abundant road supply charge negative charges to travelers arriving at them.

[0016] (Due to the clustering effect, areas with high attractiveness are prone to congestion, while areas with low attractiveness are prone to road oversupply. This invention alleviates road congestion and makes the road network traffic distribution more balanced by having attractive areas in congested areas charge positive fares to travelers arriving at those areas, and attractive areas in areas with oversupply charge negative fares to travelers arriving at those areas.)

[0017] Furthermore, the steps for establishing the lower-level model include:

[0018] S1. Construct a transportation network G(N,A), where N is the set of nodes, a is the segment number, A is the set of segment numbers, r is the origin (starting point) number, R is the set of origin (starting point) numbers, s is the destination (ending point) number, S is the set of destination (ending point) numbers, k is the path number, and K is the set of path numbers; v a Traffic flow on road segment a; Let q represent the traffic flow along path k between OD (Origin-Destination) pairs (r,s); rs Traffic flow between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0;

[0019] Traffic flow v on road segment a a satisfy:

[0020]

[0021] Traffic flow q between OD pairs (r,s) rs satisfy:

[0022]

[0023] S2. Establish the generalized travel cost function for travelers, denoted as... Broadly defined travel costs include road impedance and attraction charges. The expression is:

[0024]

[0025] Among them, t a Let be the travel time on road segment a, which is a function of the traffic flow on road segment a; β1 and β2 are unit-uniform parameters; p s The attraction fee represents the decision variable of the upper-level model (attraction operators change the generalized travel cost of travelers by setting different attraction fees, thereby affecting travelers' route choices);

[0026] S3. Establish a traveler's route selection utility function based on a stochastic user equilibrium model, denoted as U. k The traveler's path choice utility consists of two parts: fixed utility and random utility. Therefore, U k The expression is:

[0027] Uk =V k +ε k (4)

[0028]

[0029] Among them, V k The fixed utility of choosing path k for a traveler is represented by travel impedance, which is a negative number in the utility value. The larger the impedance value, the smaller the utility, and the attractive force β of the destination. 0,s and generalized travel costs The influence of ε k The stochastic utility of choosing path k for a traveler also represents the discrepancy between the traveler's perceived cost of path k and the actual cost.

[0030] Travelers are influenced by the attractiveness of destinations and the generalized cost of travel, choosing the path that maximizes their attraction and minimizes the generalized cost. According to utility maximization theory, we have:

[0031]

[0032] random utility ε k Following a Gumbel distribution, the probability that a traveler chooses path k is... Represented in the logit model as follows:

[0033]

[0034] (This shows that the number of people choosing a particular route is directly proportional to the attractiveness of the destinations connected by that route, and inversely proportional to the minimum generalized travel cost of that route and the attraction fees at those destinations.)

[0035] Therefore, the traffic flow on path k between OD pairs (r,s) is:

[0036]

[0037] S4. Considering that attraction charges affect generalized travel costs, and thus affect traffic flow between OD pairs (r,s), the traffic flow q between the OD pairs (r,s) is therefore... rs It can be expressed as a continuously monotonically decreasing function of the minimum generalized travel cost between OD pairs (r,s), i.e.:

[0038]

[0039] Where θ is the correlation coefficient of the traffic flow function between OD pairs (r,s) (i.e., the demand sensitivity parameter); u represents the maximum potential traffic flow between OD pairs (r,s); rsThe minimum generalized travel cost between OD pairs (r,s) is expressed as:

[0040]

[0041] in, It is the inverse function of the traffic flow function between OD pairs (r,s).

[0042] Furthermore, the lower-level model is the optimization model L1;

[0043] The objective function of the optimization model L1 is:

[0044]

[0045] The constraints of the optimization model L1 are:

[0046]

[0047]

[0048]

[0049]

[0050] In the above, Z(v,q) is a function of the traffic flow v on the road segment and the traffic flow q between the OD pairs (r,s); a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, S is the set of destination numbers, k is the route number, and K is the set of route numbers; v a Traffic flow on road segment a; t a Let t be the travel time on road segment a, which is a function of the traffic flow on road segment a. a (w); q rs Let β be the traffic flow between OD pairs (r,s); 0,s The attractiveness of the destination; Let be the traffic flow on path k between OD pairs (r,s); β1 and β2 are unit-uniform parameters; p s The attraction fee is represented as a decision variable in the upper-level model; It is the inverse function of the traffic flow function between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0.

[0051] Prove the model equivalence of the aforementioned lower-level model:

[0052] The Lagrangian function of model L1 is constructed as follows:

[0053]

[0054] In the above, Z is the antiderivative, and λ rs and μ a All of these are parameters introduced by the Lagrange function method;

[0055] The corresponding KKT conditions are:

[0056]

[0057]

[0058]

[0059] By rearranging (0-2) and (0-3), we can obtain:

[0060] μ a =β1t a (0-5)

[0061]

[0062] We obtain the following by summing all paths k between OD pairs (r,s) from (0-6):

[0063]

[0064] Therefore:

[0065]

[0066] From (0-6) and (0-8), we can obtain:

[0067]

[0068] Therefore, we can conclude that:

[0069]

[0070] Similarly, from (0-4), we can obtain:

[0071]

[0072] From (0-7), we can obtain:

[0073]

[0074] but:

[0075]

[0076] Right now:

[0077]

[0078] Equation (0-10) is the inverse function of elastic demand (i.e., the inverse function of traffic flow function between OD pairs (r,s)).

[0079] Since the objective function of the optimization model L1 is a strictly convex function, and considering that the constraints are linear, it is a convex set. Therefore, the optimization model L1 has a unique solution.

[0080] Furthermore, the steps for establishing the upper-level model include:

[0081] Suppose there are n destinations in the transportation network G(N,A), where there is no toll game between different destinations, all destinations belong to the same group, and all destinations pursue the maximization of common profit. In this case, the upper-level model, which aims at maximizing the profit of each destination, is equivalent to maximizing the profit of the destination's own group, and can be expressed as:

[0082] max∑ s∈s π s =∑ s∈s [W(q rs )-C(q rs (12)

[0083] W(q rs )=∑ r∈R (ω+p s )q rs (13)

[0084]

[0085] Right now:

[0086]

[0087] In the above, N is the set of nodes, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, and S is the set of destination numbers; π s To attract land profits, ∑ s∈s π s For group profits; W(q) rs C(q) represents the revenue generated from the operation of the attraction site; rs ) represents the operating costs of the attraction location (here, following Mentzer JT's assumptions, we only consider the cost of the service and the fixed costs of the attraction location, setting the production cost to 0 to simplify subsequent analysis); q rsTraffic flow between OD pairs (r,s) is obtained by solving the lower-level model; ω represents the revenue generated by each traveler reaching the destination; p s H represents the attraction fee, which is a decision variable in the upper-level model; s It is the service level of the attraction, and the cost of the attraction providing the service is... δ is the cost coefficient for the attraction to provide services to each arrival; F s These are the fixed costs of attracting visitors.

[0088] Furthermore, the steps for establishing the upper-level model include:

[0089] Suppose there are n attractions in a transportation network G(N,A), and there is a tolling game between different attractions (the tolling strategy of any attraction will be affected by the tolling strategies adopted by other attractions). All attractions belong to the same group, but each attraction seeks to maximize its own profit. In this case, the upper-level model is based on the goal of maximizing the profit of the attractions, which is equivalent to each attraction maximizing its own profit. The decision of each attraction is to choose the optimal tolling strategy to maximize its own profit. Attractions engage in non-cooperative competition, and the result of the game is that the attractions compete with each other to reach Nash equilibrium.

[0090] The profit function for attracting location s is:

[0091]

[0092] In the above, N is the set of nodes, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, and S is the set of destination numbers; W(q rs W(q) represents the revenue generated from the operation of the attraction site. rs )=∑ r∈R (ω+p s )q rs C(q) rs The cost of attracting customers is the operating cost. (Referring to Mentzer JT's assumptions, this section only considers the cost of the service and the fixed costs of the attraction location, setting the production cost to 0 to simplify subsequent analysis); q rs Traffic flow between OD pairs (r,s) is obtained by solving the lower-level model; ω represents the revenue generated by each traveler reaching the destination; p s H represents the attraction fee, which is a decision variable in the upper-level model; s It is the service level of the attraction, and the cost of the attraction providing the service is... δ is the cost coefficient for the attraction to provide services to each arrival; Fs These are the fixed costs of attracting visitors;

[0093] Let p = (p1, p2, ..., p n Let S be the vector representing all the charges for each attraction, then the set of attraction numbers is S = {1, 2, ..., n}.

[0094] When competition among attracting lands reaches Nash equilibrium, the attracting land charges satisfy the non-negativity constraint, i.e.:

[0095]

[0096] remember Let X represent the set of charging strategies for the attraction s, where X = X1 * X2 * ... * X n ;

[0097] remember The vector of all attraction charges when the attraction charges reach Nash equilibrium is the Nash equilibrium point of attraction charges, provided that equation (18) holds:

[0098]

[0099] Equation (18) indicates that when the attraction fee reaches Nash equilibrium, each attraction fee strategy is the best response to the other attraction fee strategies, and each attraction cannot increase its revenue by unilaterally changing its own attraction fee (Supplementary explanation: π). s This is the profit function of the attractor s. The upper-level model aims to maximize the attractor's own profit, which means maximizing the profit of the attractor's group. Therefore, the profit is determined when the attractor s achieves its optimal charging value. The profit must be greater than or equal to the profit when the attraction site's own charging fee does not reach the optimal value. in, This represents the charge of attraction site s at the Nash equilibrium point of attraction site charges (i.e., under the optimal attraction site charge condition); This represents the vector of charges for all attractor sites except attractor site s at the Nash equilibrium point of attractor site charges, i.e. This represents the traffic flow between OD pairs (r,s) at the Nash equilibrium point of attraction point pricing (i.e., under optimal attraction point pricing conditions), obtained by solving the lower-level model. The traffic flow on road segment a is represented by the Nash equilibrium point of the attraction toll (i.e., under the optimal attraction toll condition), which is obtained by solving the lower-level model.

[0100] From equation (16), we can see that π s It's about p sThe function, denoted here as π. s (p s Assume π s (p s ) is a continuously differentiable function on X, because For the Nash equilibrium point of attraction fees, equation (18) is equivalent to the following variational inequality:

[0101]

[0102] in,

[0103] (Additional explanation: The letters marked with "*" in the upper right corner above indicate the Nash equilibrium point for attraction pricing, i.e., the situation under optimal attraction pricing conditions.)

[0104] The present invention also provides an optimization system for attraction site charging strategy under traffic demand management, the system comprising: an information collection module, a model building module, and a model solving module;

[0105] The information collected by the information collection module includes urban road network information, historical travel distribution information, and destination information.

[0106] The model building module is used to build a two-level programming model, which includes an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of the destination, and the decision variable of the upper-level model is the destination fee. By changing the destination fee, the generalized travel cost of travelers in the lower-level model is changed, thereby affecting the travelers' route selection. The lower-level model is a traveler route selection model that considers the destination and travel cost under stochastic demand.

[0107] The model solving module solves the bi-level programming model to obtain the optimal strategy for attraction land charging under traffic demand management.

[0108] Compared with the prior art, the present invention has the following beneficial effects:

[0109] (1) This invention simplifies the management of traffic management departments. By transferring the toll collection function to the attraction area, it provides the possibility of implementing congestion pricing in more cities through a simpler toll collection method; and to facilitate implementation in more cities, the definition of attraction area in this invention is more general.

[0110] (2) This invention can effectively improve the efficiency of road transportation. Differences in attractiveness lead to uneven distribution of traffic volume on the road network, resulting in ineffective utilization of road network resources. This invention solves traffic congestion in high-attractivity areas by balancing the attractiveness of each destination to travelers through tolling, and can greatly improve the operational efficiency of the transportation network.

[0111] (3) This invention can increase public support. If travelers are charged directly, it will cause dissatisfaction among travelers with traffic management; however, this invention greatly reduces the dissatisfaction caused by direct charges by indirectly charging travelers for congestion.

[0112] (4) In the literature considering the attractiveness of destinations to travelers, attractiveness models are mainly used for the location of shopping malls and logistics centers, and the attractiveness of transportation and shipping services to travelers, with almost no consideration given to the impact of destination attractiveness on travelers' route choices. This invention innovatively proposes a route choice model that considers destination attractiveness, which more accurately describes the impact of attractiveness pricing methods on travelers.

[0113] (5) This invention also specifically explores attraction pricing under two market mechanisms: monopoly market and monopolistic competition. It describes optimization methods and systems for attraction pricing strategies under the two markets, providing traffic managers under different operating systems with more diversified pricing options.

[0114] A monopolistic market is a market structure where there is only one supplier and numerous demanders. In this case, it is assumed that all destinations belong to the same group, which seeks to maximize their collective benefit. Therefore, a function maximizing total benefit is established in the upper-level model. For travelers, considering the positive utility of reaching the target destination and the negative utility of the journey, they choose the path that maximizes their own benefit; therefore, a stochastic user equilibrium model is proposed to describe the traveler's path selection, and an equivalent optimization model is constructed.

[0115] Monopolistic competition refers to a market with many suppliers selling similar but dissimilar products. In reality, for destination operators, it's more common for destinations to be operated by multiple independent companies. These different operating companies compete with each other, and each company aims to maximize its own revenue. Furthermore, any company's strategy is influenced by the strategies adopted by other companies. Therefore, this is essentially an oligopolistic market. In a monopolistic competition market, travelers' route selection follows the same principles as in a monopolistic market; however, in the higher-level model, destinations, pursuing revenue maximization, engage in toll-charging games with each other, and each company's goal is to maximize its own revenue. Again, any company's strategy is influenced by the strategies adopted by other companies. Ultimately, an equilibrium pricing is reached, meaning that a destination cannot increase its revenue by increasing or decreasing its own price.

[0116] The principles involved in this invention include:

[0117] Nash equilibrium theory is an effective tool for analyzing non-cooperative game problems in economics and management practice; it is also known as non-cooperative game equilibrium. In a game, regardless of the other players' strategies, one player will always choose a certain strategy; this strategy is called the dominant strategy. If any player's chosen strategy is optimal given that the strategies of all other players are fixed, then this combination is defined as a Nash equilibrium.

[0118] Consider the K-person non-cooperative Nash equilibrium problem:

[0119] Let convex subset Let u be the strategy set of player i, and let u be the utility function of player i. i :X i →R, where X=X1*X 2x *…*X K Let I = {1, 2, ..., K}, x -i =(x1,…,x i-1 ,x i+1 ,…,x K ), x = (x1, x2, ..., x K )=(x i ,x -i ), X -i =X1*…*X i-1 *X i+1 *…*X K .

[0120] Theorem 1 This is called the Nash equilibrium point if the following equation holds:

[0121]

[0122] Suppose that for any i ∈ I, the function u i It is continuously differentiable on X. If If x is the Nash equilibrium point, then * It is also a solution to the following variational inequalities:

[0123]

[0124] in, F(x * ) is the function F(x) at the Nash equilibrium point x. * The value of F(x) at time * ) T F(x) * The transpose of ).

[0125] Furthermore, for any i ∈ I and any given x -i ∈X -i function ui (x i ,x -i Regarding x s If it is pseudoconvex, then the variational inequality is a necessary condition.

[0126] The present invention will be further described in detail below with reference to specific embodiments and accompanying drawings, but this does not imply any limitation on the scope of protection of the present invention. Attached Figure Description

[0127] Figure 1 This is a system framework diagram of the optimized attraction land charging strategy under traffic demand management according to Embodiments 1 and 2 of the present invention.

[0128] Figure 2 This is a schematic diagram of the Nguyen-Dupuis network in Embodiment 1 and Embodiment 2 of the present invention.

[0129] Figure 3 This is the convergence graph of the genetic algorithm in Embodiment 1 of the present invention.

[0130] Figure 4 This is a schematic diagram of the congestion situation before toll collection in Embodiment 1 of the present invention.

[0131] Figure 5 This is a schematic diagram of congestion after toll collection in Embodiment 1 of the present invention.

[0132] Figure 6 This is a comparison chart of traffic flow on the road section before and after toll collection in Embodiment 1 of the present invention.

[0133] Figure 7 This is a comparison chart of travel time on the road section before and after toll collection in Embodiment 1 of the present invention.

[0134] Figure 8 This is a comparison chart of the attraction revenue before and after charging fees in Embodiment 1 of the present invention.

[0135] Figure 9 This is a graph showing the relationship between the revenue from attracting land and the demand sensitivity parameter before and after charging in Embodiment 1 of the present invention.

[0136] Figure 10 This is the convergence graph of the iterative heuristic algorithm in Embodiment 2 of the present invention.

[0137] Figure 11 This is a schematic diagram of the congestion situation before toll collection in Embodiment 2 of the present invention.

[0138] Figure 12 This is a schematic diagram of congestion after toll collection in Embodiment 2 of the present invention.

[0139] Figure 13 This is a comparison chart of traffic flow on the road section before and after toll collection in Embodiment 2 of the present invention.

[0140] Figure 14 This is a comparison chart of travel time on the road section before and after toll collection in Embodiment 2 of the present invention.

[0141] Figure 15 This is a comparison chart of the optimal charging rates for attracting tourists in Embodiment 1 and Embodiment 2 of the present invention.

[0142] Figure 16 This is a comparison chart of the land attraction benefits of Embodiment 1 and Embodiment 2 of the present invention.

[0143] Figure 17 This is a comparison chart of the total travel time of the road network in Embodiment 1 and Embodiment 2 of the present invention. Detailed Implementation

[0144] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0145] Example 1

[0146] This example presents an optimization method and system for attraction-based pricing strategies under traffic demand management.

[0147] The method includes:

[0148] A two-level programming model is established, comprising an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of all destinations in a city (in a monopolistic market, all destinations in the city aim to maximize the common profit). The decision variable of the upper-level model is the destination fee, which changes the generalized travel cost of travelers in the lower-level model, thereby influencing travelers' route choices. The lower-level model is a traveler route selection model considering both destinations and travel costs under stochastic demand.

[0149] Solving the bilevel programming model yields the optimal strategy for attraction pricing under traffic demand management.

[0150] The steps for establishing the lower-level model in this example include:

[0151] S1. Construct a transportation network G(N,A), where N is the set of nodes, a is the segment number, A is the set of segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, S is the set of destination numbers, k is the path number, and K is the set of path numbers; v a Traffic flow on road segment a; Let q represent the traffic flow on path k between OD pairs (r,s); rs Traffic flow between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0;

[0152] Traffic flow v on road segment a a satisfy:

[0153]

[0154] Traffic flow q between OD pairs (r,s) rs satisfy:

[0155]

[0156] S2. Establish the generalized travel cost function for travelers, denoted as... Broadly defined travel costs include road impedance and attraction charges. The expression is:

[0157]

[0158] Among them, t a Let be the travel time on road segment a, which is a function of the traffic flow on road segment a; β1 and β2 are unit-uniform parameters; p s The attraction fee is represented as a decision variable in the upper-level model;

[0159] S3. Establish a traveler's route selection utility function based on a stochastic user equilibrium model, denoted as U. k The traveler's path choice utility consists of two parts: fixed utility and random utility. Therefore, U k The expression is:

[0160] U k =V k +ε k (4)

[0161]

[0162] Among them, V k The fixed utility of choosing path k for a traveler is represented by travel impedance, which is a negative number in the utility value. The larger the impedance value, the smaller the utility, and the attractive force β of the destination. 0,s and generalized travel costs The influence of ε k The stochastic utility of choosing path k for a traveler also represents the discrepancy between the traveler's perceived cost of path k and the actual cost.

[0163] Travelers are influenced by the attractiveness of destinations and the generalized cost of travel, choosing the path that maximizes their attraction and minimizes the generalized cost. According to utility maximization theory, we have:

[0164]

[0165] random utility ε k Following a Gumbel distribution, the probability that a traveler chooses path k is... Represented in the logit model as follows:

[0166]

[0167] Therefore, the traffic flow on path k between OD pairs (r,s) is:

[0168]

[0169] S4. Considering that attraction charges affect generalized travel costs, and thus affect traffic flow between OD pairs (r,s), the traffic flow q between the OD pairs (r,s) is therefore... rs It can be expressed as a continuously monotonically decreasing function of the minimum generalized travel cost between OD pairs (r,s), i.e.:

[0170]

[0171] Where θ is the correlation coefficient of the traffic flow function between OD pairs (r,s); u represents the maximum potential traffic flow between OD pairs (r,s); rs The minimum generalized travel cost between OD pairs (r,s) is expressed as:

[0172]

[0173] in, It is the inverse function of the traffic flow function between OD pairs (r,s).

[0174] The lower-level model described in this example is the optimization model L1.

[0175] The objective function of the optimization model L1 is:

[0176]

[0177] The constraints of the optimization model L1 are:

[0178]

[0179]

[0180]

[0181]

[0182] In the above, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, S is the set of destination numbers, k is the route number, and K is the set of route numbers; v a Traffic flow on road segment a; t a Let t be the travel time on road segment a, which is a function of the traffic flow on road segment a. a (w); q rs Let β be the traffic flow between OD pairs (r,s); 0,s The attractiveness of the destination; Let be the traffic flow on path k between OD pairs (r,s); β1 and β2 are unit-uniform parameters; p s The attraction fee is represented as a decision variable in the upper-level model; It is the inverse function of the traffic flow function between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0.

[0183] The steps for establishing the upper-level model in this example include:

[0184] Suppose there are n destinations in the transportation network G(N,A), where there is no toll game between different destinations, all destinations belong to the same group, and all destinations pursue the maximization of common profit. In this case, the upper-level model, which aims at maximizing the profit of each destination, is equivalent to maximizing the profit of the destination's own group, and can be expressed as:

[0185] max∑ s∈s π s =∑ s∈S [W(q rs )-C(q rs (12)

[0186] W(q rs )=∑ r∈R (ω+p s )q rs (13)

[0187]

[0188] Right now:

[0189]

[0190] In the above, N is the set of nodes, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, and S is the set of destination numbers; π s To attract land profits, ∑ s∈s π s For group profits; W(q) rs C(q) represents the revenue generated from the operation of the attraction site; rs ) is the operating cost of the attraction site; q rs Traffic flow between OD pairs (r,s) is obtained by solving the lower-level model; ω represents the revenue generated by each traveler reaching the destination; p s H represents the attraction fee, which is a decision variable in the upper-level model; s It is the service level of the attraction, and the cost of the attraction providing the service is... δ is the cost coefficient for the attraction to provide services to each arrival; F s These are the fixed costs of attracting visitors.

[0191] The system includes: an information acquisition module, a model building module, and a model solving module;

[0192] The information collected by the information collection module includes urban road network information, historical travel distribution information, and destination information.

[0193] The model building module is used to build the bi-level programming model in this example method. The bi-level programming model includes an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of the destination. The decision variable of the upper-level model is the destination fee. By changing the destination fee, the generalized travel cost of travelers in the lower-level model is changed, thereby affecting the travelers' route selection. The lower-level model is a traveler route selection model that considers the destination and travel cost under stochastic demand.

[0194] The model solving module solves the bi-level programming model to obtain the optimal strategy for attraction land charging under traffic demand management.

[0195] The system framework diagram for optimizing the attraction pricing strategy under traffic demand management in this example is as follows: Figure 1 As shown.

[0196] To test the effectiveness of the proposed optimization method and system for attraction-based pricing strategy under traffic demand management, the following approach is used: Figure 2 The classic Nguyen-Dupuis network is shown in the example analysis. Figure 2The table lists two origins (Origins 1 and 4) and two destinations (Destination 2 and 3) with the same function (assuming they are both hospitals or shopping malls, etc.). The destinations include both their own parking spaces and nearby roadside parking spaces. Segment numbers, free-flow travel times, and segment capacities are shown in Table 1. The table also includes information on potential demand for origins (ODs). They are respectively Assuming that attraction point 3 has a higher level of service and a better location, it is more attractive to travelers. Given β 0,2 =1,β 0,3 =2, H2=15, H3=20, and the corresponding attraction destination 3 has higher fixed operating costs, therefore F2=5000, F3=10000. Assume the deviation between travelers' perceived and actual travel costs for path k is a fixed value, i.e., ε. k =ε=0.001. Other parameters: β1=0.9, β2=0.6, θ=0.025, ω=100, δ=0.2.

[0197] Table 1. Road segment number, free-flow travel time, and road segment capacity

[0198]

[0199]

[0200] In the genetic algorithm, the initial population size is M = 20, the crossover probability is Pc = 0.75, the mutation probability is Pm = 0.05, and the maximum number of generations is g = 100. The initial cost p0 for all attractors is 0.

[0201] For attraction fees in a monopolistic market, MATLAB is used for programming calculations, and the iterative process of the genetic algorithm is as follows: Figure 3 As shown, the algorithm converges around generation 30. Setting the fees for the two attraction sites (attraction site 2 and attraction site 3) to 9.21 (yuan / vehicle) and 33.00 (yuan / vehicle) respectively can maximize the total revenue of the two attraction sites.

[0202] The traffic operation evaluation indicators are shown in Table 2.

[0203] Table 2 Traffic Operation Evaluation Indicators

[0204]

[0205] From Table 2 TTI index and Figure 4 , Figure 5It can be seen that, for attractions 2 and 3 with the same function, travelers are more willing to choose attraction 3 when it is not charged, because attraction 3 has better service quality and higher attractiveness. However, they are not aware of the negative externalities they are generating on the transportation network. At this time, road segments 13 and 19 connecting attraction 3 are severely congested, and road segment 16 is moderately congested. By implementing differentiated pricing, attraction 3 charges higher fees to control traffic demand and reduce negative externalities such as traffic congestion. The traffic flow and travel time on road segments 13, 16, and 19 connecting attraction 3 decrease most significantly (see...). Figure 6 and Figure 7 On section 16, traffic flow decreased by 39.47%, and travel time decreased by 41.25%, changing the section's operational status from moderate congestion to smooth flow. On sections 13 and 19, traffic flow decreased by 16.11%, and travel time decreased by 29.88%, changing from severe congestion before toll collection to basically smooth flow. For attraction 2, there was basically no congestion in the surrounding roads before toll collection, but the road operation was not at its optimal state. After charging a small fee, the surrounding roads were able to achieve smooth flow.

[0206] Overall, before the toll was implemented, the road network had 2 severely congested sections, 1 moderately congested section, and only 8 uncongested sections; after the toll was implemented, there were no severely congested sections, and 15 uncongested sections. Figure 17 It can be seen that the total travel time of the road network decreased from 85163.38 min to 59763.03 min, a reduction of 29.83%.

[0207] Depend on Figure 6 , Figure 7 It can be seen that both traffic flow and travel time decreased after the implementation of attraction tolls. Table 3 compares the demand for private cars before and after the implementation of attraction tolls. After the implementation of attraction tolls, the demand for private car travel decreased by 326, a decrease of 16.82%. This indicates that charging for attraction tolls significantly affects its demand, and that traffic demand management through attraction tolls is effective.

[0208] Table 3. Changes in Odone Demand Before and After Attraction Fees

[0209]

[0210] Depend on Figure 8 and Figure 16 It can be seen that the maximum total revenue after charging fees for the two attractions is 12.942102 × 10 4 Yuan, before charging was 11.770055 × 10 4 The total revenue after charging was 9.96% higher than before charging. For a single attraction, the revenue of attraction 2 before charging was 6.833606 × 10. 4 Yuan, after charging, it is 7.194407 × 104 Yuan; the revenue before charging for attraction site 3 was 4.936449 × 10 4 The original price was 5.747695 yuan, which, after the fee, is 10. 4 The revenue of both attractions increased compared to before the toll was 5.28% and 16.43%, respectively. Therefore, attraction operators are more willing to cooperate with traffic managers in traffic demand management.

[0211] Figure 9 The relationship between the revenue from attracting land and the demand sensitivity parameter θ before and after charging is described. As shown in the figure, during the change of the demand sensitivity parameter θ, the revenue from attracting land after charging is always greater than or equal to that before charging, and the difference between the two decreases as the demand sensitivity parameter θ increases.

[0212] In summary, under a monopolistic market, charging according to the optimal strategy of attraction-based pricing in this example has significant effects on traffic flow regulation and traffic demand management. At the same time, the revenue of the attraction will not be affected by the implementation of this pricing strategy. Therefore, this method and system are feasible at the levels of traffic managers, attraction sites, and travelers.

[0213] Example 2

[0214] This example presents an optimization method and system for attraction-based pricing strategies under traffic demand management.

[0215] The method includes:

[0216] A two-level programming model is established, comprising an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of each destination (in a monopolistic competition market (i.e., an oligopolistic market), each destination in the city pursues its own profit maximization). The decision variable of the upper-level model is the destination fee, which changes the generalized travel cost of travelers in the lower-level model, thereby influencing travelers' route choices. The lower-level model is a traveler route selection model considering both destinations and travel costs under stochastic demand.

[0217] Solving the bilevel programming model yields the optimal strategy for attraction pricing under traffic demand management.

[0218] The steps for establishing the lower-level model in this example include:

[0219] S1. Construct a transportation network G(N,A), where N is the set of nodes, a is the segment number, A is the set of segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, S is the set of destination numbers, k is the path number, and K is the set of path numbers; v a Traffic flow on road segment a; Let q represent the traffic flow on path k between OD pairs (r,s); rs Traffic flow between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0;

[0220] Traffic flow v on road segment a a satisfy:

[0221]

[0222] Traffic flow q between OD pairs (r,s) rs satisfy:

[0223]

[0224] S2. Establish the generalized travel cost function for travelers, denoted as... Broadly defined travel costs include road impedance and attraction charges. The expression is:

[0225]

[0226] Among them, t a Let be the travel time on road segment a, which is a function of the traffic flow on road segment a; β1 and β2 are unit-uniform parameters; p s The attraction fee is represented as a decision variable in the upper-level model;

[0227] S3. Establish a traveler's route selection utility function based on a stochastic user equilibrium model, denoted as U. k The traveler's path choice utility consists of two parts: fixed utility and random utility. Therefore, U k The expression is:

[0228] U k =V k +ε k (4)

[0229]

[0230] Among them, V k The fixed utility of choosing path k for a traveler is represented by travel impedance, which is a negative number in the utility value. The larger the impedance value, the smaller the utility, and the attractive force β of the destination. 0,s and generalized travel costs The influence of ε k The stochastic utility of choosing path k for a traveler also represents the discrepancy between the traveler's perceived cost of path k and the actual cost.

[0231] Travelers are influenced by the attractiveness of destinations and the generalized cost of travel, choosing the path that maximizes their attraction and minimizes the generalized cost. According to utility maximization theory, we have:

[0232]

[0233] random utility ε k Following a Gumbel distribution, the probability that a traveler chooses path k is... Represented in the logit model as follows:

[0234]

[0235] Therefore, the traffic flow on path k between OD pairs (r,s) is:

[0236]

[0237] S4. Considering that attraction charges affect generalized travel costs, and thus affect traffic flow between OD pairs (r,s), the traffic flow q between the OD pairs (r,s) is therefore... rs It can be expressed as a continuously monotonically decreasing function of the minimum generalized travel cost between OD pairs (r,s), i.e.:

[0238]

[0239] Where θ is the correlation coefficient of the traffic flow function between OD pairs (r,s); u represents the maximum potential traffic flow between OD pairs (r,s); rs The minimum generalized travel cost between OD pairs (r,s) is expressed as:

[0240]

[0241] in, It is the inverse function of the traffic flow function between OD pairs (r,s).

[0242] The lower-level model described in this example is the optimization model L1.

[0243] The objective function of the optimization model L1 is:

[0244]

[0245] The constraints of the optimization model L1 are:

[0246]

[0247]

[0248]

[0249]

[0250] In the above, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, S is the set of destination numbers, k is the route number, and K is the set of route numbers; v a Traffic flow on road segment a; t a Let t be the travel time on road segment a, which is a function of the traffic flow on road segment a. a (w); q rs Let β be the traffic flow between OD pairs (r,s); 0,s The attractiveness of the destination; Let be the traffic flow on path k between OD pairs (r,s); β1 and β2 are unit-uniform parameters; p s The attraction fee is represented as a decision variable in the upper-level model; It is the inverse function of the traffic flow function between OD pairs (r,s); For variables of 0-1, when road segment a between OD pairs (r,s) is on path k, The value is 1, otherwise The value is 0.

[0251] The steps for establishing the upper-level model in this example include:

[0252] Suppose there are n attractions in a transportation network G(N,A), and there is a tolling game between different attractions. All attractions belong to the same group, but each attraction seeks to maximize its own profit. In this case, the upper-level model is based on the goal of maximizing the profit of the attractions, which is equivalent to maximizing the profit of each attraction. The decision of each attraction is to choose the optimal tolling method to maximize its own profit. The attractions compete non-cooperatively, and the result of the game is that the attractions compete with each other to reach Nash equilibrium.

[0253] The profit function for attracting location s is:

[0254]

[0255] In the above, N is the set of nodes, a is the road segment number, A is the set of road segment numbers, r is the origin number, R is the set of origin numbers, s is the destination number, and S is the set of destination numbers; W(q rs W(q) represents the revenue generated from the operation of the attraction site. rs )=∑ r∈R (ω+p s )qrs C(q) rs The cost of attracting customers is the operating cost. q rs Traffic flow between OD pairs (r,s) is obtained by solving the lower-level model; ω represents the revenue generated by each traveler reaching the destination; p s H represents the attraction fee, which is a decision variable in the upper-level model; s It is the service level of the attraction, and the cost of the attraction providing the service is... δ is the cost coefficient for the attraction to provide services to each arrival; F s These are the fixed costs of attracting visitors;

[0256] Let p = (p1, p2, ..., p n Let S be the vector representing all the charges for attracting locations, then S = {1, 2, ..., n};

[0257] When competition among attracting lands reaches Nash equilibrium, the attracting land charges satisfy the non-negativity constraint, i.e.:

[0258]

[0259] remember Let X represent the set of charging strategies for the attraction s, where X = X1 * X2 * ... * X n ;

[0260] remember The vector of all attraction charges when the attraction charges reach Nash equilibrium is the Nash equilibrium point of attraction charges, provided that equation (18) holds:

[0261]

[0262] Equation (18) indicates that when attraction pricing reaches Nash equilibrium, each attraction pricing strategy is the best response to the pricing strategies of other attractions, and each attraction cannot increase its revenue by unilaterally changing its own attraction pricing; where, This represents the charge of attraction site s at the Nash equilibrium point of attraction site charges (i.e., under the optimal attraction site charge condition); This represents the vector of charges for all attractor sites except attractor site s at the Nash equilibrium point of attractor site charges, i.e. This represents the traffic flow between OD pairs (r,s) at the Nash equilibrium point of attraction point pricing (i.e., under optimal attraction point pricing conditions), obtained by solving the lower-level model. The traffic flow on road segment a is represented by the Nash equilibrium point of the attraction toll (i.e., under the optimal attraction toll condition), which is obtained by solving the lower-level model.

[0263] From equation (16), we can see that π s It's about p s The function, denoted here as π. s (p s Assume π s (p s ) is a continuously differentiable function on X, because For the Nash equilibrium point of attraction fees, equation (18) is equivalent to the following variational inequality:

[0264]

[0265] in,

[0266] The system includes: an information acquisition module, a model building module, and a model solving module;

[0267] The information collected by the information collection module includes urban road network information, historical travel distribution information, and destination information.

[0268] The model building module is used to build the bi-level programming model in this example method. The bi-level programming model includes an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of the destination. The decision variable of the upper-level model is the destination fee. By changing the destination fee, the generalized travel cost of travelers in the lower-level model is changed, thereby affecting the travelers' route selection. The lower-level model is a traveler route selection model that considers the destination and travel cost under stochastic demand.

[0269] The model solving module solves the bi-level programming model to obtain the optimal strategy for attraction land charging under traffic demand management.

[0270] The system framework diagram for optimizing the attraction pricing strategy under traffic demand management in this example is as follows: Figure 1 As shown.

[0271] To test the effectiveness of the proposed optimization method and system for attraction-based pricing strategy under traffic demand management, the following approach is used: Figure 2 The classic Nguyen-Dupuis network is shown in the example analysis. Figure 2 The table lists two origins (Origins 1 and 4) and two destinations (Destination 2 and 3) with the same function (assuming they are both hospitals or shopping malls, etc.). The destinations include both their own parking spaces and nearby roadside parking spaces. Segment numbers, free-flow travel times, and segment capacities are shown in Table 1. The table also includes information on potential demand for origins (ODs). They are respectively Assuming that attraction point 3 has a higher level of service and a better location, it is more attractive to travelers. Given β 0,2 =1,β 0,3 =2, H2=15, H3=20, and the corresponding attraction destination 3 has higher fixed operating costs, therefore F2=5000, F3=10000. Assume the deviation between travelers' perceived and actual travel costs for path k is a fixed value, i.e., ε. k =ε=0.001. Other parameters: β1=0.9, β2=0.6, θ=0.025, ω=100, δ=0.2.

[0272] Figure 10 The convergence graph of an iterative heuristic algorithm under oligopolistic market conditions is described. Figure 10 As can be seen, the equilibrium solution is obtained around the 6th generation in this example, which shows that the proposed algorithm is effective.

[0273] From Table 2 TTI index and Figure 11 It can be seen that, for attractions 2 and 3 with the same function, travelers are more willing to choose attraction 3 when it is not charged, because attraction 3 has better service quality and higher attractiveness. However, they are not aware of the negative externalities they generate on the transportation network. At this time, road segments 13 and 19 connecting attraction 3 are severely congested, and road segment 16 is moderately congested. Under an oligopolistic market, there are 12 smooth road segments after attraction charges are implemented (see...). Figure 12 This reduces the number of scenarios by three compared to the monopolistic market scenario in Example 1.

[0274] Depend on Figure 13 , Figure 14 It can be seen that under the oligopolistic market, the traffic flow and travel time of the congested road sections near attraction 3 were greatly improved after the toll was implemented. The travel time of road sections 13, 16 and 19, which are directly connected to attraction 3, decreased by the largest percentage, at 25.75%, 38.12% and 25.75%, respectively.

[0275] A comparative analysis of Example 1 and Example 2 is conducted:

[0276] like Figure 15 , 16 As shown in Example 1, the optimal charges for attractions 2 and 3 under a monopolistic market are 9.21 yuan / vehicle and 33.00 yuan / vehicle, respectively, with revenues of 7.194407 × 10⁻⁶. 4 Yuan, 5.747695×10 4 In Example 2, under an oligopolistic market, without cooperation, the charges for attractions 2 and 3 are 4.83 yuan / vehicle and 27.61 yuan / vehicle respectively, with revenues of 7.163750 × 10⁻⁶. 4 Yuan, 5.727624×104 Yuan; In Example 2, the tolls and revenues under the oligopolistic market are lower than those under the monopolistic market in Example 1, and the traffic demand is also slightly higher than that under the monopolistic market in Example 1 (see Table 4). Therefore, even after toll collection, two road sections still experience mild congestion (see Table 4). Figure 12 However, in this case, neither party in the oligopolistic market of Example 2 can increase their revenue by unilaterally increasing or decreasing their own charges (i.e., achieving Pareto optimality), and the total travel time of the road network (see...) Figure 17 ) and attraction revenue (see Figure 16 Both scenarios are superior to the situation where no fees are charged. When the attractors in the oligopolistic market scenario of Example 2 cooperate and simultaneously increase their fees, the revenue of both parties can be increased.

[0277] Table 4 Comparison of Private Car Demand in Monopolistic and Oligopolistic Markets

[0278]

[0279] Examples 1 and 2 demonstrate that, without considering public transportation, the optimal tolling strategy employed in this example significantly improves traffic flow regulation and demand management, while ensuring that attraction revenue is not negatively impacted. Therefore, this method and system are feasible at the levels of traffic managers, attraction locations, and travelers, alleviating concerns about reduced revenue and reluctance to use attraction tolling.

[0280] Regarding congestion pricing, most current research and existing implementation experience focus on congestion pricing based on road sections or warning lines. While this method is effective, it is difficult to implement and has low feasibility. This invention innovatively proposes a pricing strategy based on attraction sites, aiming to provide more cities with new measures to alleviate road congestion. Examples 1 and 2 quantitatively analyze the effects of attraction site pricing on alleviating traffic congestion and improving social welfare. By comparing the improvements in road congestion and social welfare before and after the implementation of pricing, the effectiveness of attraction site pricing is demonstrated.

[0281] In summary, in both monopolistic and oligopolistic markets, the optimal toll collection strategy according to this invention plays a crucial role in traffic flow regulation and demand management. Furthermore, the implementation of this invention does not harm the profits of the destination, and the method and system can be implemented at the traffic manager, destination, and traveler levels. This invention does not consider public transportation; if public transportation were taken into account, the currently reduced demand for private cars would shift to public transportation to reach the same destination. Therefore, the number of people reaching the destination would be higher than when only private cars were considered, and the destination's profits would be higher than the current results.

[0282] The present invention has been described above by way of example in conjunction with the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the embodiments shown herein.

Claims

1. An optimization method for attraction-based pricing strategies under traffic demand management, characterized in that, The method includes: A two-level programming model is established, comprising an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of the destination, and the decision variable of the upper-level model is the destination fee. By changing the destination fee, the generalized travel cost of travelers in the lower-level model is altered, thereby influencing travelers' route selection. The lower-level model is a traveler route selection model that considers both the destination and travel cost under stochastic demand. Solving the bilevel programming model yields the optimal strategy for attraction pricing under traffic demand management; The term "attraction" refers to facilities that are attractive to urban residents and tourists; The attraction fee refers to the charging measures taken by an attraction to travelers arriving at that attraction, including positive and negative fees. Attractions in areas with road congestion charge travelers arriving at that attraction with positive fees, while attractions in areas with road oversupply charge travelers arriving at that attraction with negative fees. The steps for establishing the lower-level model include: S1, Constructing a transportation network , For a set of nodes, For road segment numbering, A set of road segment numbers, Number the departure point. A set of departure point numbers, Numbering the attraction sites For the set of attraction site numbers, For path numbering, A set of path numbers; For road section Traffic flow on the road; for right Inter-path Traffic flow on the road; for right Traffic flow between; For 0-1 variables, when right Intersection In the path When, The value is 1, otherwise The value is 0; Section Traffic flow satisfy: ； (1) right Traffic flow between satisfy: ； (2) S2. Establish the generalized travel cost function for travelers, denoted as... Broadly defined travel costs include road impedance and attraction charges. The expression is: ， (3) in, For road section The travel time on the road segment A function of traffic flow; , Standardize parameters for each unit; The attraction fee is represented as a decision variable in the upper-level model; S3. Establish a traveler's route selection utility function based on a stochastic user equilibrium model, denoted as... The traveler's path choice utility consists of two parts: fixed utility and random utility. The expression is: ， (4) ， (5) in, Selecting routes for travelers The fixed utility of travel impedance is expressed as a negative value in the utility value; the greater the impedance value, the smaller the utility, and the greater the attraction of the attracted place. and generalized travel costs The impact; Selecting routes for travelers The random utility also represents the traveler's preference for the route. The discrepancy between the perceived and actual cost of travel; Travelers are influenced by the attractiveness of destinations and the generalized cost of travel, choosing the path that maximizes their attraction and minimizes the generalized cost. According to utility maximization theory, we have: ， ； (6) random utility Following a Gumbel distribution, travelers choose paths. probability Represented in the logit model as follows: ； (7) therefore right Inter-path Traffic flow on the road is as follows: ； (8) S4. Considering that attraction fees affect the broader cost of travel, and thus influence... right Traffic flow between, therefore, the aforementioned right Traffic flow between Represented as right The minimum generalized travel cost between [a certain number of] is a continuously monotonically decreasing function, i.e.: ， (9) in, for right The correlation coefficient of traffic flow functions between them; for right The maximum potential traffic flow between them; For OD The minimum generalized travel cost between them is expressed as: ， (10) in, ,for right The inverse function of the traffic flow function.

2. The optimization method for attraction-based pricing strategy under traffic demand management according to claim 1, characterized in that, The lower-level model is the optimization model L1; The objective function of the optimization model L1 is: (11) The constraints of the optimization model L1 are: ， (11a) ， (11b) ， (11c) ； (11d) In the above, For road segment numbering, A set of road segment numbers, Number the departure point. A set of departure point numbers, Numbering the attraction sites For the set of attraction site numbers, For path numbering, A set of path numbers; For road section Traffic flow on the road; For road section The travel time on the road segment The function of traffic flow, denoted here as ; for right Traffic flow between; The attractiveness of the destination; for right Inter-path Traffic flow on the road; , Standardize parameters for each unit; The attraction fee is represented as a decision variable in the upper-level model; for right The inverse function of the traffic flow function; For 0-1 variables, when right Intersection In the path When, The value is 1, otherwise The value is 0.

3. The optimization method for attraction-based pricing strategy under traffic demand management according to any one of claims 1-2, characterized in that, The steps for establishing the upper-level model include: Assuming in the transportation network There exists In a scenario where there is no toll game between different attractions, all attractions belong to the same group, and all attractions pursue the maximization of common profits, the upper-level model, where the goal of each attraction maximizing its own profit is the same as the goal of each attraction maximizing the profits of its own group, is expressed as: ， (12) ， (13) ， (14) Right now: ； (15) In the above, For a set of nodes, For road segment numbering, A set of road segment numbers, Number the departure point. A set of departure point numbers, Numbering the attraction sites A set of attraction site numbers; To attract land profits, For group profits; W( (This refers to the revenue generated from the operation of the attraction site.) It is the operating cost of the attraction; for right Traffic flow between them is obtained by solving the lower-level model; The revenue that each traveler brings to the destination; The attraction fee is represented as a decision variable in the upper-level model; It is the service level of the attraction, and the cost of the attraction providing the service is... , It is the cost factor for an attraction to provide services to each arrival; These are the fixed costs of attracting visitors.

4. An optimization method for an attraction-based pricing strategy under traffic demand management according to any one of claims 1-2, characterized in that, The steps for establishing the upper-level model include: Assuming in the transportation network There exists There are several attractions, and there is a toll game between different attractions. All attractions belong to the same group, but each attraction pursues its own profit maximization. In this case, the upper-level model is based on the goal of maximizing the profit of each attraction. The decision of each attraction is to choose the optimal attraction to charge to maximize its own profit. Attractions compete non-cooperatively, and the result of the game is that the attractions compete with each other to reach Nash equilibrium. Attraction The profit function is: ， (16) In the above, For a set of nodes, For road segment numbering, A set of road segment numbers, Number the departure point. A set of departure point numbers, Numbering the attraction sites A set of attraction site numbers; W( (This refers to the revenue generated from attracting tourists to the area.) ; It is the operating cost of attracting customers. ; for right Traffic flow between them is obtained by solving the lower-level model; The revenue that each traveler brings to the destination; The attraction fee is represented as a decision variable in the upper-level model; It is the service level of the attraction, and the cost of the attraction providing the service is... , It is the cost factor for an attraction to provide services to each arrival; These are the fixed costs of attracting visitors; remember Let represent the vector of all attraction charges, then the set of attraction numbers. ; When competition among attracting lands reaches Nash equilibrium, the attracting land charges satisfy the non-negativity constraint, i.e.: ； (17) remember , indicating attractiveness The set of charging strategies, ; remember , which represents the vector of all attraction charges when the attraction charges reach Nash equilibrium, is the Nash equilibrium point of attraction charges, provided that equation (18) holds: ； (18) Equation (18) indicates that when attraction pricing reaches Nash equilibrium, each attraction pricing strategy is the best response to the pricing strategies of other attractions, and each attraction cannot increase its revenue by unilaterally changing its own attraction pricing; where, This represents the attraction point at which the attraction fee is calculated. The fee; This indicates the Nash equilibrium point at which attraction fees are calculated, excluding attraction fees. The vector of all attraction charges outside, i.e. ; This indicates the Nash equilibrium point for attracting land charges. right Traffic flow between them is obtained by solving the lower-level model; This indicates the road segment at the Nash equilibrium point for attraction pricing. The traffic flow is obtained by solving the lower-level model. From equation (16), we can see that It is about The function, denoted here as ; Assumption yes A continuously differentiable function on, because For the Nash equilibrium point of attraction fees, equation (18) is equivalent to the following variational inequality: ， (19) in, .

5. An optimization system for attraction-based pricing strategies under traffic demand management, characterized in that, The system adopts an optimization method for attraction-based charging strategy under traffic demand management as described in any one of claims 1-4. The system includes: an information collection module, a model building module, and a model solving module. The information collected by the information collection module includes urban road network information, historical travel distribution information, and destination information. The model building module is used to build a two-level programming model, which includes an upper-level model and a lower-level model. The upper-level model aims to maximize the profit of the destination, and the decision variable of the upper-level model is the destination fee. By changing the destination fee, the generalized travel cost of travelers in the lower-level model is changed, thereby affecting the travelers' route selection. The lower-level model is a traveler route selection model that considers the destination and travel cost under stochastic demand. The model solving module solves the bi-level programming model to obtain the optimal strategy for attraction land charging under traffic demand management.

Images