Dynamic pricing method in high-speed rail presale period based on passenger flow distribution

Abstract

The invention relates to the technical field of traffic, and discloses a passenger flow distribution-based dynamic pricing method in a high-speed rail pre-sale period, which comprises the following steps of: aiming at a ticket price floating scene in the pre-sale period, constructing a multi-dimensional ticket price adjustment rule strategy set comprising a line, a train and a ticket selling date, establishing a differentiated dual-target pricing model in the pre-sale period, according to the method, factors such as economy, convenience, rapidity and comfort are considered, a passenger travel fixed cost formula and a path-related cost formula are provided for two scenes of a single-line or non-overlapping section road network and an overlapping section road network, and a passenger flow distribution dynamic adjustment strategy and a matched passenger flow distribution optimization method are designed to carry out reasonable passenger flow distribution; according to the method, reasonable distribution of passenger flow is realized, and the feasibility and scientificity of dynamic pricing in the high-speed rail presale period are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transportation, and particularly to a dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution. Background Art

[0002] In the process of the market-oriented operation of high-speed rail, formulating a scientific fare dynamic decision-making mechanism to achieve the effective allocation of transportation resources to balance the market supply and demand relationship has become an important development trend in high-speed rail operation. A floating mechanism is adopted outside the pre-sale period, while inside the pre-sale period, the ticket quota allocation plan is adjusted according to the remaining ticket situation to adapt to the supply and demand changes. The following problems mainly exist: (1) In terms of dynamic pricing, existing theories have deficiencies in the research and differential treatment of fare mechanisms inside and outside the pre-sale period, lacking an overall strategy and adjustment mechanism that can be optimized to adapt to large-scale complex road networks and fare fluctuations during the pre-sale period; (2) In terms of passenger flow demand, currently widely used demand functions (such as elastic demand function, time-varying demand function, ticket purchase intensity function, etc.) divide the daily passenger flow during the pre-sale period by the travel probability (sparse first and then dense or equally spaced), lacking the transfer situation of ticket purchase demand during the pre-sale period (such as the refund and change behavior of passengers); (3) In terms of passenger flow distribution, relying on idealized models easily leads to a significant deviation between the passenger flow distribution and the actual situation (such as the passenger flow being overly concentrated in a certain period or a certain train, and some passenger flows cannot be allocated to get on the train), lacking a passenger flow distribution adjustment strategy and method under game competition. Summary of the Invention

[0003] Aiming at the above deficiencies in the prior art, the present invention provides a dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution, which is used to solve the problems existing in the existing fare dynamic decision-making methods, such as being unable to adapt to large-scale complex road networks, low operating efficiency, low corporate benefits or high passenger travel costs, and being unable to reasonably distribute passenger flows.

[0004] In order to achieve the above invention purpose, the technical solution adopted by the present invention is as follows:

[0005] A dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution includes the following steps:

[0006] S1. Obtain OD passenger flow data and train timetable data, conduct path search, screen reasonable service paths, and calculate path capacity;

[0007] S2. Based on the passenger ticket purchase historical data of the target line and the number of days until departure, divide the time of the ticket pre-sale period into time zones to generate several time slices;

[0008] S3. Based on the station level and OD passenger flow, and in accordance with the ticket selling order during the pre-sale period, conduct OD level division for each time slice to generate several OD levels for each time slice, and sort the OD levels according to the priority level from high to low to generate sorted OD levels;

[0009] S4. Based on the sorted OD levels, in the order of priority from high to low, the same OD levels are successively segmented and randomly arranged to generate several OD passenger flow blocks of the same OD level;

[0010] S5. Randomly select an OD passenger flow block of the current OD level and calculate the generalized travel costs of passengers on different paths under the current OD passenger flow;

[0011] S6. Construct a bi-objective function that maximizes the enterprise revenue and minimizes the passenger travel cost. After setting the constraint conditions, solve it to obtain the selection probabilities of the current OD passenger flow block for different paths;

[0012] S7. Based on the selection probabilities of the current OD passenger flow block for different paths and the remaining capacity of the train, update the actual boarding number;

[0013] S8. Based on the updated actual boarding number, judge whether to adjust the fare. If so, execute step S6. Otherwise, judge whether all OD passenger flow blocks have been traversed. If so, execute step S9. Otherwise, execute step S5;

[0014] S9. Judge whether all OD levels have been traversed. If so, execute step S10. Otherwise, execute step S4;

[0015] S10. Judge whether all time slices have been traversed. If so, output the passenger flow distribution result. Otherwise, execute step S3.

[0016] The present invention has the following beneficial effects:

[0017] 1. A dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution proposed by the present invention combines the train serialization and airline revenue management ideas, constructs a multi-dimensional fare adjustment rule strategy set that adapts to large-scale complex road networks including lines, trains, and ticket sales dates, and can flexibly respond to market demand changes while maintaining operational efficiency;

[0018] 2. Considering the refund and change behaviors of passengers during the pre-sale period and the transfer demand of passenger flow between different dates, coupling the realistic constraints of railway transportation organization, a differential bi-objective pricing model during the pre-sale period is proposed, aiming to balance the enterprise revenue and the passenger travel cost;

[0019] 3. Taking whether there are overlapping line sections as the judgment criterion, using the multinomial Logit model and the path length Logit model to describe the fixed travel cost and the comprehensive travel cost of passengers respectively. When dealing with the problem that the deviation between the traditional passenger flow distribution method and the actual operation situation is large, a dynamic adjustment strategy for passenger flow distribution and an optimization method based on the dynamic adjustment strategy are proposed, so as to realize the reasonable distribution of passenger flow and finally improve the feasibility and scientificity of dynamic pricing during the pre-sale period of high-speed rail. Brief Description of the Drawings

[0020] Figure 1 It is a schematic flow chart of a dynamic pricing method during the advance sale period of high-speed trains based on passenger flow distribution proposed by the present invention;

[0021] Figure 2 It is a schematic diagram of the dynamic adjustment process of passenger flow distribution in the embodiment. Detailed Embodiment

[0022] The following describes the detailed embodiment of the present invention to facilitate those skilled in the art of this technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the detailed embodiment. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions created using the concept of the present invention are within the scope of protection.

[0023] As Figure 1 shown, a dynamic pricing method during the advance sale period of high-speed trains based on passenger flow distribution includes the following steps S1 - S10:

[0024] S1. Obtain OD passenger flow data and train timetable data, search for direct paths and one-transfer paths, screen reasonable service paths, and calculate path capacities.

[0025] In this embodiment, OD passenger flow data and train timetable data are selected and used as input data to search for direct paths and one-transfer paths, so as to screen reasonable service paths, eliminate unreasonable service paths and service paths with a capacity of 0, and at the same time add reasonable service paths to the attributes of the corresponding OD; in addition, the purpose of calculating path capacity is to clarify the maximum passenger flow carrying capacity of different paths. The direct path capacity is equal to the seating capacity of a single train, and the transfer path capacity is equal to the minimum value of the seating capacities of multiple trains. Among them, OD is Origin - Destination, that is, the origin and destination of a traffic trip or the origin and destination of passenger flow.

[0026] S2. Based on the passenger ticket purchase history data of the target line, obtain the time characteristics of passenger ticket purchase behavior, and divide the advance sale period of passenger tickets into time zones according to the number of days before departure to generate several time slices.

[0027] In this embodiment, it is mainly the execution stage of the pre-sale period slicing strategy. According to the passenger ticket purchase historical data of the target line, the passenger ticket pre-sale period is sliced to facilitate the selection of the passenger flow in each slice according to the time sequence. The principle is as follows: According to the historical ticket data (passenger ticket purchase historical data), the time characteristics of the ticket purchase behavior of high-speed rail passengers are obvious, that is, when the departure date is far away, the daily passenger ticket purchase volume is low; when the departure date is close, especially within 3 days before departure, the daily passenger ticket purchase volume surges, and the peak of the ticket purchase data appears on the second day before the departure date, with a strong periodic pattern. Therefore, considering the differences in the ticket purchase behavior of passengers during the pre-sale period, the pre-sale period is divided into N time slices. The ticket purchase behaviors of passengers in different time slices are heterogeneous, and the ticket purchase behaviors of passengers within a single time slice are homogeneous. Different fare adjustment strategies can be adopted for each time slice. At the same time, in the pre-sale period slicing strategy, the number of time slices should not be too small, otherwise the differences in the ticket purchase behavior of passengers cannot be reflected, nor should it be too large. Excessive subdivision of the ticket purchase behavior of passengers will lead to too high complexity and affect the algorithm efficiency of passenger flow distribution. It is generally appropriate to take N as 3 to 5. Among them, the pre-sale period slicing is as Figure 2 shown, Figure 2 The first part of Figure 2 shows the principle and process of time slicing, that is, N time slices are set according to the number of days before departure. The first time slice is the farthest from the departure time, followed by the second time slice, and so on. The Nth time slice is the closest to the departure time.

[0028] S3. Based on the station level and OD passenger flow, according to the pre-sale period ticket sales order, conduct OD level division for each time slice to generate several OD levels for each time slice, and sort the OD levels according to the priority order from high to low to generate the sorted OD levels.

[0029] In this embodiment, it is mainly the execution stage of the OD stratification strategy. According to factors such as station level and OD passenger flow volume, the OD level is divided and adjusted to select hierarchical OD passenger flows from large to small. The principle is as follows: Based on the station classification in the train dimension of the multi-dimensional fare floating strategy set, combined with factors such as the daily average OD passenger flow volume and the daily average OD turnover volume, the OD level is divided; since the coincidence degree of high-level OD and long-distance OD is relatively high, high-level OD is preferentially allocated; at the same time, the alternative set of single-door trains is screened in advance to preferentially allocate single-door OD; among them, the multi-dimensional fare floating strategy set includes the line dimension, the train dimension, and the ticket sales date dimension; (1) Line dimension: The economic development level and passenger income level in different regions are different, and the acceptance of fares is different; for different lines, different line basic fare rates are set based on historical ticket data, and different fare floating ranges are set based on the line basic fare rates; currently, the freight rate of the second-class seat on China's high-speed rail is generally between 0.4 and 0.5 yuan per person-kilometer, and the fare ratio of business class, first class, and second class is approximately 3:1.6:1; the higher the economic development level of the areas passed by the line, the higher the average passenger income level, and the more tense the supply and demand situation, the higher the line basic fare rate, and the more significant the fare floating effect; (2) Train dimension: In the train dimension, indicators such as the administrative level of urban nodes, the development level of urban nodes, the status and function of urban nodes in the high-speed rail network, the daily average passenger flow arrivals and departures at stations, the daily average number of trains served by stations, and the layout of high-speed rail maintenance productivity are comprehensively considered to divide the station level; based on the number of stations served by trains and the station level, combined with factors such as the average passenger flow volume of trains, the average turnover volume of trains, the average passenger mileage, and the unit distance yield rate of trains, a train value pedigree is constructed, the train level is set, different types of trains are set with different fare adjustment ranges, and the same type of trains are set with different gear fare adjustment ranges according to influencing factors; it should be noted that the train value pedigree is an inductive analysis of trains according to their operation characteristics, and the operation efficiency of trains is affected by multiple factors such as the number of served OD, OD demand, the number of trains, and the train capacity, which does not mean that the operation efficiency of high-speed trains with fewer stops must be higher than that of medium- and low-speed trains with more stops; (3) Ticket sales date dimension: In the ticket sales date dimension, the idea of airline revenue management is introduced. For the situation of supply exceeding demand (such as the average occupancy rate of the line being relatively low or the occupancy rate of the train section being relatively low), based on the time interval between the ticket sales date and the departure date, different time periods are set with fare reduction ranges or reduction ratios, and the earlier the ticket purchase time, the greater the fare discount; for the situation of supply falling short of demand, the fare reduction in the ticket sales date dimension is not considered.

[0030] S4. Based on the sorted OD level, in the order of priority from high to low, the same OD level is successively blocked and randomly arranged to generate several OD passenger flow blocks of the same OD level.

[0031] In this embodiment, it is mainly the execution stage of the block disordering strategy. According to the OD level (OD grade), the passenger flow OD is processed in blocks. The size of the passenger flow block is related to the OD level and is arranged in disorder to randomly select the OD passenger flow block as the object of passenger flow allocation. The principle is as follows: Blocking means restricting the maximum number of passengers boarding each time within the OD passenger flow demand and train capacity. The size of the passenger flow block is related to the passenger flow level to avoid the problem that a certain level (grade) or a certain OD occupies too much capacity of the same train. Disordering means that for the same-level OD (same-grade OD), disordered allocation is carried out in units of passenger flow blocks to avoid the problem that some passengers cannot board due to the late boarding order. Among them, Figure 2 The second part of Figure 2 shows the OD grading process, that is, OD grading is carried out according to the selected time slice, and at the same time, the priorities of OD grades are divided. Those with higher priorities are sorted to the front and divided into I-IV grades in turn. Figure 2 The third part of Figure 2 shows the OD grade disordered block process, that is, disordered block operations are carried out on the same OD grades in turn according to the order of priority levels, keeping the same-grade OD internally divided into multiple passenger flow blocks and arranged in disorder, and setting the maximum number of people in each OD passenger flow block.

[0032] S5. Randomly select an OD passenger flow block of the current OD grade and calculate the generalized travel costs of passengers on different paths under the current OD passenger flow.

[0033] In this embodiment, it is mainly the calculation stage of the path travel cost, which is to calculate the generalized travel costs of passengers on different paths under the current OD passenger flow. Among them, the generalized travel costs of passengers include the calculation of the fixed costs of the single-line or non-overlapping section road network and the comprehensive costs of the overlapping section road network. Specifically:

[0034] (1) Single-line or non-overlapping section road network: For a single line or multiple lines without overlapping sections, a fixed travel cost formula for passengers is constructed from aspects such as fare economy, time rapidity, transfer convenience, and seat comfort, and the MNL model is used for vehicle-flow matching. The calculation formula is:

[0035]

[0036] Among them, represents the fixed travel cost of passengers in the t pre-sale period belonging to OD w choosing seat s on path k. α, β, and γ respectively represent the time cost parameter, fare cost parameter, and seat comfort cost parameter, and their sum is 1. represents the in-transit travel time of passengers in the t pre-sale period belonging to OD w choosing path k. ψ represents the transfer penalty coefficient, with the unit of hour / time. represents the number of transfers of passengers in the t pre-sale period belonging to OD w choosing path k. Denote the ticket price for seat s on path k selected by passengers whose pre-sale period t belongs to OD w, in yuan, κ w Denote the value of time of passengers for OD w, in yuan per hour, Denote the comfort cost for seat s on path k selected by passengers whose pre-sale period t belongs to OD w.

[0037] In addition, since a passenger's journey may involve multiple trains, the above formula needs to be comprehensively evaluated for multiple trains. That is, for fare economy, the total fare of the passenger's journey is considered; for time efficiency, the total running time of the trains is considered; for transfer convenience, the penalty time needs to be considered; for the comfort of the seat, the average comfort of multiple trains is taken. At the same time, all items are uniformly converted into time cost for calculation.

[0038] (2) Road network with overlapping sections: For multiple lines with overlapping sections, since path overlap violates the irrelevant independence assumption of the MNL model, the Path Size Logit (PSL) model is used instead of the MNL. The generalized cost formula for passenger travel is as follows:

[0039]

[0040] Among them, Denote the comprehensive travel cost for seat s on path k selected by passengers whose pre-sale period t belongs to OD w, Denote the path-related cost for path k selected by passengers whose pre-sale period belongs to OD w, μ represents the fitting parameter, and μ > 0. If there is a lack of data for fitting, 1 can be taken. ln represents the logarithmic function, Denote the length of path section g during pre-sale period t, in kilometers, Denote the total length of path k during pre-sale period t, in kilometers, R t Denote the set of all reasonable service paths during pre-sale period t, λ r,g Denote a 0-1 variable, which is 1 if path r contains path section g, otherwise 0.

[0041] S6. Construct a bi-objective function that maximizes the enterprise revenue and minimizes the passenger travel cost, set the constraint conditions, and at the same time use the multinomial Logit model or the path length Logit model for solution to calculate the selection probabilities of the current OD passenger flow for different paths.

[0042] In this embodiment, a bi-objective function that maximizes the enterprise revenue and minimizes the passenger travel cost is constructed, and the constraint conditions are set for the establishment of the differential bi-objective pricing model and constraint conditions during the pre-sale period, specifically as follows:

[0043] (1) Objective function: To study the issue of fare fluctuations, the following assumptions are made: 1) The research object only includes high-speed railways, without considering other transportation modes. 2) Train overbooking is not considered in the research scenario. 3) Each passenger is a rational person, and their travel demands are independent of each other. Among them, for enterprises, Z1 is used to represent the total revenue of high-speed rail transportation, indicating the optimization goal related to enterprise revenue, that is:

[0044]

[0045] Among them, max means to take the maximum value, represents the passenger flow of seat s on path k selected by passengers in OD w during the t pre-sale period, represents the fare of seat s on path k selected by passengers in OD w during the t pre-sale period, τ(t) represents the refund probability during the t pre-sale period, and z(t) represents the refund handling fee rate during the t pre-sale period.

[0046] For passengers, Z2 is used to represent the total travel cost of passengers, indicating the optimization goal related to passenger benefits, that is:

[0047]

[0048] Among them, min means to take the minimum value, represents the travel cost of seat s on path k selected by passengers in OD w during the t pre-sale period, and according to different application scenarios, or

[0049] (2) Constraint conditions:

[0050] 1) Refund and rescheduling constraints during the pre-sale period. Since a passenger's rescheduling behavior can be split into a refund behavior and a new ticket purchase behavior, the refund-related functions can directly describe the refund behavior of passengers during the high-speed rail pre-sale period and indirectly describe the rescheduling behavior of passengers during the high-speed rail pre-sale period. Therefore, both the refund handling fee rate function z(t) and the refund probability function τ(t) are related to the pre-sale date, that is:

[0051]

[0052] τ(t) = 1 - e -λt

[0053] Among them, T represents the high-speed rail pre-sale period, in days, t = T is the departure day, t = T - 1 is the day before departure, e represents the exponential function, and λ represents the passenger refund parameter, λ > 0.

[0054] 2) Fare fluctuation constraints. Fare fluctuation constraints are used to limit the reasonable fluctuation range of high-speed rail fare rates; too high a fare rate is not conducive to passenger travel, and too low a fare rate is not conducive to enterprise revenue, that is:

[0055]

[0056] Among them, represents the initial fare of seat s on path k selected by passengers with OD w. δ represents the maximum increase and decrease amplitudes of the fare respectively.

[0057] 3) Fare non-inversion constraint. Since the transportation cost of short-distance passengers is lower than that of long-distance passengers, the short-distance fare should be lower than the long-distance fare. To prevent passengers from avoiding the long-distance fare by purchasing multi-segment short-distance tickets, when the travel distance is the same, the sum of short-distance fares should not be lower than the long-distance fare, that is:

[0058]

[0059] Among them, n represents the number of train sections of OD w on path k. represents the fare of seat s on path k selected by OD passengers corresponding to the sc-th train section during the t pre-sale period.

[0060] 4) OD passenger flow constraint. The sum of the passenger flows assigned to each reasonable service path for a single OD should not be greater than the passenger flow demand of that OD, and the sum of the assigned passenger flows of all routes should not be greater than the total passenger flow demand of all ODs, that is:

[0061]

[0062] Among them, q w represents the passenger flow travel demand of OD w, and Q represents the total passenger flow travel demand of all ODs.

[0063] 5) Path passenger flow non-negativity constraint. According to common sense, the assigned passenger flow of each reasonable service path cannot be negative, that is:

[0064]

[0065] 6) Path capacity constraint. The path capacity constraint is used to limit the maximum passenger flow capacity of each reasonable service path, which is limited by the minimum value of the seating capacities of all trains on the space-time service path, that is:

[0066]

[0067] Among them, represents a 0-1 variable, which takes 1 if the k-th path of OD w contains the space-time service arc segment (i, j), and 0 otherwise. Cap i,j represents the capacity of the space-time service arc segment (i, j), and its value is the sum of the capacities of all trains serving this arc segment.

[0068] 7) Train connection time constraint. In the transfer path, when two trains are connected at the transfer station, the arrival time of the previous train should be earlier than that of the next train, i.e.:

[0069] dep(tr2,st)>arr(tr1,st)

[0070] Where dep(tr2,st) represents the departure time of the next train tr2 at the transfer station st, and arr(tr1,st) represents the arrival time of the previous train tr1 at the transfer station st.

[0071] 8) Transfer times constraint. In the transfer path, there is an upper limit to the maximum acceptable number of transfers for passengers, and it varies according to different regions, different lines, and different income levels, i.e.:

[0072] u ≤ U

[0073] Where u represents the number of transfers of the passenger, and U represents the maximum acceptable number of transfers for the passenger.

[0074] 9) Transfer time constraint. In the transfer path, sufficient time needs to be reserved for passengers to perform transfer operations at the transfer station, i.e.:

[0075] e st ≥ E st

[0076] Where e st represents the transfer time of the passenger at the transfer station st, and E st represents the minimum transfer time at the transfer station st.

[0077] In addition, during the traditional vehicle-flow matching process, the number of passengers boarding each time depends on the OD passenger flow demand and the train capacity, taking the minimum value of the two. However, in the actual boarding process, different OD passenger flows board the train mixedly. In the passenger flow distribution algorithm, different OD passenger flows board the train in sequence, which may lead to the following problems: First, due to the limited train capacity, only a certain level, one or several OD pairs are served; Second, due to the sequential upstream of vehicle-flow matching, some passenger flow OD pairs are ranked behind, and there are cases where some passengers do not board the train. If these OD pairs are high-level OD pairs or large passenger flow OD pairs, the deviation from the actual situation is greater; Finally, the exclusive train (that is, a certain OD pair or some OD pairs can only meet the travel demand through a certain train) is not considered. If the capacity of the exclusive train is occupied, the exclusive OD passenger flow has no train to board. Therefore, in order to reduce the gap between the passenger flow distribution result and the actual situation, a dynamic adjustment strategy and optimization algorithm for passenger flow distribution are proposed from aspects such as presale period slicing, OD stratification, block disorder, and vehicle-flow matching. Among them, presale period slicing, OD stratification, and block disorder have been introduced in the above steps S2 - S5. Here, the main introduction is to construct a double-objective function of maximizing enterprise revenue and minimizing passenger travel costs, set constraint conditions, and at the same time use the Multi-Nominal Logit (MNL) model or the Path Size Logit (PSL) model for solution, so as to perform vehicle-flow matching. The principle of vehicle-flow matching is as follows: According to the random utility theory, each choice of a passenger has a corresponding utility value, which includes two parts: the observable part and the unobservable part (usually called the random error term). When facing multiple selectable paths, assuming that the random error term follows the Gumbel distribution, calculate the utility functions of the observable parts of different paths, which respectively represent the probabilities of passengers choosing each path. The commonly used models are the multi-nominal logit model and its derivative models. At the same time, combined with the generalized cost formula of passenger travel, for two scenarios of a single-line or non-overlapping section road network and an overlapping section road network, the multi-nominal logit model and the path length logit model are respectively applicable for vehicle-flow matching, that is:

[0078]

[0079]

[0080] Among them, represents the multi-nominal logit selection probability that a passenger in the presale period t belonging to OD pair w chooses seat s on path k, represents the path length logit selection probability that a passenger in the presale period t belonging to OD pair w chooses seat s on path k. exp represents the exponential function, and θ1 and θ2 respectively represent the multi-nominal logit and path length logit selection probability parameters.

[0081] S7. Determine whether the remaining capacity of the train is greater than or equal to the OD passenger flow demand multiplied by the path selection probability. If so, update the actual boarding number to the OD passenger flow demand multiplied by the path selection probability; otherwise, update the actual boarding number to the remaining capacity of the train, and generate the updated actual boarding number.

[0082] S8. Based on the updated actual boarding number, determine whether to adjust the fare. If so, execute step S6; otherwise, determine whether all OD passenger flow blocks have been traversed. If so, execute step S9; otherwise, execute step S5.

[0083] S9. Determine whether all OD levels have been traversed. If so, execute step S10; otherwise, execute step S4.

[0084] S10. Determine whether all time periods have been traversed. If so, output the passenger flow distribution result; otherwise, execute step S3.

[0085] In summary, a dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution proposed by the present invention first combines the train pedigree and airline revenue management ideas to construct a multi-dimensional fare adjustment rule strategy set that adapts to large-scale complex road networks including lines, trains, and ticket sales dates, and flexibly responds to market demand changes while maintaining operational efficiency. At the same time, considering the refund and change behaviors of passengers during the pre-sale period and the transfer demand of passenger flow between different dates, and coupling the realistic constraints of railway transportation organization, a differential double-objective pricing model during the pre-sale period is proposed to balance the enterprise revenue and the passenger travel cost. Finally, taking whether there are overlapping line sections as the judgment criterion, using the multinomial Logit model and the path length Logit model to describe the fixed cost of passenger travel and the comprehensive cost of passenger travel respectively, in response to the problem that the deviation between the traditional passenger flow distribution method and the actual operation situation is relatively large, a dynamic adjustment strategy for passenger flow distribution and an optimization method based on the dynamic adjustment strategy are proposed to realize the reasonable distribution of passenger flow, and ultimately improve the feasibility and scientificity of dynamic pricing during the pre-sale period of high-speed rail.

[0086] Specific embodiments are applied in the present invention to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention. At the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0087] Those of ordinary skill in the art will realize that the embodiments described herein are provided to assist the reader in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention based on these technical revelations disclosed in the present invention, and these deformations and combinations are still within the scope of protection of the present invention.

Claims

1. A dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution, characterized in that It includes the following steps: S1. Obtain OD passenger flow data and train timetable data, conduct path search, screen reasonable service paths, and calculate path capacity; S2. Based on the passenger ticket purchase history data of the target line and the number of days until departure, conduct time partitioning of the ticket pre-sale period to generate several time slices; S3. Based on the station level and OD passenger flow, conduct OD level division for each time slice according to the ticket selling order during the pre-sale period, generate several OD levels for each time slice, and sort the OD levels according to the priority order from high to low to generate the sorted OD levels; S4. Based on the sorted OD levels, in the order of priority from high to low, block and randomly arrange the same OD levels in turn to generate several OD passenger flow blocks of the same OD level; S5. Randomly select an OD passenger flow block of the current OD level and calculate the generalized travel costs of passengers on different paths under the current OD passenger flow; S6. Construct a bi-objective function of maximizing enterprise revenue and minimizing passenger travel costs, set constraints and then solve it to obtain the selection probabilities of the current OD passenger flow block for different paths; S7. Update the actual boarding number based on the selection probabilities of the current OD passenger flow block for different paths and the remaining capacity of the train; S8. Based on the updated actual boarding number, judge whether to adjust the ticket price. If so, execute step S6. Otherwise, judge whether all OD passenger flow blocks have been traversed. If so, execute step S9. Otherwise, execute step S5; S9. Judge whether all OD levels have been traversed. If so, execute step S10. Otherwise, execute step S4; S10. Judge whether all time slices have been traversed. If so, output the passenger flow distribution result. Otherwise, execute step S3.

2. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 1, wherein In step S1, the path search is direct path search and one-transfer path search.

3. The dynamic pricing method during the advance ticket sales period of high-speed rail based on passenger flow distribution according to claim 2, wherein If the path search is direct path search, the path capacity is equal to the seating capacity of a single train. If the path search is one-transfer path search, the path capacity is equal to the minimum value of the seating capacities of multiple trains.

4. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 3, characterized in that, Step S2 specifically includes: Based on the passenger ticket purchase history data of the target line, obtain the time characteristics of passenger ticket purchase behavior; Based on the time characteristics of passenger ticket purchase behavior and the number of days until departure, conduct time partitioning of the ticket pre-sale period to generate N time slices; Among them, the first time slice is the farthest from the departure time, followed by the second time slice, and so on. The Nth time slice is the closest to the departure time.

5. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 4, characterized in that In step S5, the generalized travel costs of passengers on different paths under the current OD passenger flow include the fixed travel cost of a single-line or non-overlapping section road network and the comprehensive travel cost of an overlapping section road network.

6. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 5, wherein The calculation formula for the fixed travel cost of a single-line or non-overlapping section road network is: Among them, represents the fixed travel cost for passengers in the t pre-sale period belonging to ODw to choose seat s on route k. α, β, and γ respectively represent the time cost parameter, fare cost parameter, and seat comfort cost parameter. represents the in-transit travel time for passengers in the t pre-sale period belonging to ODw to choose route k. ψ represents the transfer penalty coefficient. represents the number of transfers for passengers in the t pre-sale period belonging to ODw to choose route k. represents the fare for passengers in the t pre-sale period belonging to ODw to choose seat s on route k. κ w represents the time value of passengers of ODw. represents the comfort cost for passengers in the t pre-sale period belonging to ODw to choose seat s on route k.

7. The dynamic pricing method during the advance sale period of high-speed rail based on passenger flow distribution according to claim 6, wherein The calculation formula for the comprehensive travel cost of an overlapping section road network is: Among them, represents the comprehensive travel cost of passengers whose pre-sale period belongs to ODw for seat s on route k, represents the path-related cost of passengers whose pre-sale period belongs to ODw for route k. μ represents the fitting parameter, and ln represents the logarithmic function, represents the length of path segment g during pre-sale period t, represents the total length of route k during pre-sale period t, R t represents the set of all reasonable service routes during pre-sale period t, λ r,f represents a 0-1 variable.

8. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 7, wherein Step S6 specifically includes: S601. Construct a bi-objective function of maximizing enterprise revenue and minimizing passenger travel costs, that is: Among them, Z1 represents the objective function of maximizing enterprise revenue, and max represents taking the maximum value. represents the passenger flow of seat s on path k selected by passengers belonging to ODw during the t pre-sale period. represents the ticket price of seat s on path k selected by passengers belonging to ODw during the t pre-sale period, τ(t) represents the ticket refund probability during the t pre-sale period, z(t) represents the ticket refund handling fee rate during the t pre-sale period, Z2 represents the objective function of minimizing passengers' travel costs, and min represents taking the minimum value. represents the travel cost of seat s on path k selected by passengers belonging to ODw during the t pre-sale period, and ∨ represents "or". S602. Construct the pre-sale period refund and change constraints, that is: Among them, z represents the refund handling fee rate function, τ represents the refund probability function, T represents the advance sale period of high-speed trains, t = T is the departure day, t = T - 1 is the day before departure, e represents the exponential function, and λ represents the passenger refund parameter; S603. Construct the fare floating constraint, that is: Among them, represents the initial fare of seat s on path k selected by passengers of ODw, δ represents the maximum increase and decrease amplitudes of the fare respectively; S604. Construct the fare non-reversal constraint, that is: where n represents the number of train sections of ODw on path k, represents the fare of seat s on path k selected by OD passengers for the sc-th train section during the t pre-sale period; S605. Construct the OD passenger flow constraint, that is: where q w represents the passenger flow travel demand of ODw, and Q represents the total passenger flow travel demand of all OD; S606. Construct the non-negative path passenger flow constraint, that is: S607. Construct the path capacity constraint, that is: Among them, represents a 0-1 variable, Cap i,j represents the capacity of the spatio-temporal service arc segment (i, j); S608. Construct the train connection time constraint, that is: dep(tr2, st) > arr(tr1, st) Among them, dep(tr2, st) represents the departure time of the subsequent train tr2 at the transfer station st, and arr(tr1, st) represents the arrival time of the previous train tr1 at the transfer station st; S609. Construct the transfer times constraint, that is: u ≤ U Among them, u represents the number of passenger transfers, and U represents the maximum acceptable number of passenger transfers; S610. Construct the transfer time constraint, that is: e st ≥E st Among them, e st represents the transfer time of passengers at the transfer station st, and E st represents the minimum transfer time at the transfer station st; S611. Use the multinomial Logit model or the path length Logit model for solution to calculate the selection probabilities of the current OD passenger flow for different paths.

9. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 8, characterized in that, The formula for calculating the selection probabilities of the current OD passenger flow for different paths in step S611 is: Among them, represents the multinomial Logit choice probability that a passenger whose pre-sale period is t and belongs to ODw selects seat s on route k. represents the path length Logit choice probability that a passenger whose pre-sale period is t and belongs to ODw selects seat s on route k. exp represents the exponential function, and θ1 and θ2 respectively represent the multinomial Logit and path length Logit choice probability parameters.

10. The dynamic pricing method during the pre-sale period of high-speed rail based on passenger flow distribution according to claim 9, characterized in that, Step S7 specifically includes: Based on the selection probabilities of the current OD passenger flow block for different paths and the remaining train capacity, judge whether the remaining train capacity is greater than or equal to the OD passenger flow demand multiplied by the path selection probability. If so, update the actual boarding number to the OD passenger flow demand multiplied by the path selection probability. Otherwise, update the actual boarding number to the remaining train capacity, and finally generate the updated actual boarding number.

Images