A personalized travel plan sequence generation method based on passenger travel characteristics

Abstract

The application discloses a kind of personalized travel scheme sequence generation methods based on passenger travel characteristics, mainly including the following steps: establishing the identification method of passenger travel characteristics, calibrating the travel preference of ticketing passenger, calculating the travel preference weight of ticketing passenger, collecting multi-mode travel scheme ticket information, generating the five major steps of personalized multi-mode travel scheme sequence.The application is based on the historical travel characteristics of passenger, fully considers the preference difference of each passenger to multi-mode travel scheme attribute, realizes a kind of personalized multi-mode travel scheme sequence generation method, provides guidance suggestion for improving the personalized matching degree of travel service platform travel scheme, to improve the service quality of online travel ticketing service.

Description

Technical Field

[0001] This invention relates to the field of transportation, and more particularly to a method for generating personalized travel plan sequences based on passenger travel characteristics. Background Technology

[0002] With the rapid popularization of online ticketing in the transportation sector, people's demand for travel service platforms that provide online ticketing is gradually developing towards personalization and diversification. Given the contradiction between diverse travel needs and relatively fixed transport capacity, some intercity travel routes often experience ticket shortages. In such situations, travel service platforms recommend alternative travel options to passengers; however, these options often suffer from low personalization and limited travel modes.

[0003] Existing travel plan generation technologies, in terms of personalized recommendation services, only categorize users into groups such as time-sensitive and price-sensitive travelers. They fail to quantitatively discuss the relationship between traveler preferences and travel decisions, leading to similar recommended plans for different users and low personalization. Users still need to spend significant time and effort searching and comparing ticketing information to select a travel plan that better meets their expectations. This reduces users' ticket purchasing efficiency and increases the operational burden on the platform.

[0004] While current online travel ticketing systems have integrated ticketing information for various modes of transportation, personalized ticketing services remain insufficient. Therefore, it is necessary to develop a personalized multi-modal travel plan sequence generation technology to generate more tailored travel plans based on each passenger's travel preferences. Summary of the Invention

[0005] Purpose of the invention: The purpose of this invention is to provide a method for generating personalized travel plan sequences based on passenger travel characteristics that can improve the matching degree between personalized travel plans and passengers, enhance users' ticket purchasing efficiency, and achieve a win-win situation for both users and the platform.

[0006] Technical solution: The personalized travel plan sequence generation method of the present invention includes the following steps:

[0007] S1 collects the historical travel records of N passengers from the travel service platform, where passenger i has N travel records. i Then, determine whether the ticket-purchasing passenger i has a historical travel record in the travel service platform, and calculate the travel characteristics of passenger i according to the travel characteristic recognition method;

[0008] S2, combining passenger travel characteristics, divides the travel preference factors influencing passenger ticketing behavior into convenience preference D. i Efficiency Preference E i and economic preference Fi Then, based on the travel characteristics of passenger i, the travel preferences of passenger i are identified.

[0009] S3. Using the analytic hierarchy process (AHP), the relative importance of each pair of travel preferences is calculated to construct a preference discrimination matrix. The weight vector W = {ω1, ω2, ω3} of travel preferences is calculated, where ω1, ω2, and ω3 are the weights of convenience preference, efficiency preference, and economy preference, respectively.

[0010] S4, based on the ticket-purchasing passenger's target route Y, obtain multi-mode travel plans for the route; wherein, the ticketing attribute information of the j-th travel plan for the target route Y includes: the total travel time H of the travel plan. Yj Ticket Price I Yj and the transfer time G of the transfer scheme Yj ;

[0011] S5, standardize the ticketing attribute information of the multi-modal travel plan, and then calculate the optimal solution Z of the target route Y travel plan. + And the worst solution Z - And sequentially calculate the outgoing travel options for route Y and the optimal solution Z. + And the worst solution Z - The Euclidean distance between them is calculated; then, the TOPSIS method is used, combined with the weight vector W={ω1,ω2,ω3} of travel preferences, to sort the multimodal travel plans and generate a personalized travel plan sequence for ticket-purchasing passengers.

[0012] Further, in step S1, historical travel records of N passengers are collected, and the travel characteristics of passenger i's j-th historical trip are calculated; the travel characteristics include: average transfer ratio A. ij Average time ratio B ij and average price ratio C ij Their expressions are as follows:

[0013] A ij =TT ij / TT Aij

[0014] B ij =IT ij / IT Aij

[0015] C ij =P ij / P Aij

[0016] In the formula, TT ij TT represents the transfer time for passenger i's j-th historical trip; AijThis represents the maximum historical transfer time for all passengers traveling to the same origin and destination. If there is no transfer, this parameter is skipped. ij IT represents the total travel time of passenger i's j-th historical trip. Aij P represents the maximum historical total travel time for all passengers originating from and destined for the same destination; ij P represents the ticket price for passenger i's j-th historical trip. Aij This represents the maximum historical travel fare for all passengers traveling to the same origin and destination.

[0017] Furthermore, in step S1, it is determined whether the ticket-purchasing passenger i has a historical travel record in the travel service platform. The determination principle is as follows:

[0018] If the judgment result is true, and passenger i has N i If there are 10 historical travel records, then calculate N. i The average travel characteristics of each historical trip are used as the travel characteristics of the ticket-purchasing passenger i, and their expressions are as follows:

[0019]

[0020]

[0021] If the judgment result is false, the average value of all passenger travel characteristics is calculated as the travel characteristic of the ticket-purchasing passenger i, and the expressions are as follows:

[0022]

[0023] In the formula, A N B N C N These represent the average of the average transfer rate, average time rate, and average price rate for all passengers, respectively.

[0024] Furthermore, in step S2, the process of identifying passenger travel preferences based on the travel characteristics of the ticket-purchasing passenger is as follows:

[0025] S21, based on the travel characteristics of passenger i, the travel preference factors influencing passenger i's ticket purchasing behavior are divided into convenience preference D. i Efficiency Preference E i and economic preference F i Their expressions are as follows:

[0026]

[0027] When passenger i has a travel history on the travel service platform, the values ​​of A, B, and C are respectively: A i B i C i ;

[0028] When the ticket-purchasing passenger i has no historical travel record in the travel service platform, the values ​​of A, B, and C are respectively: A i ′、B i ′、C i ′.

[0029] Furthermore, in step S3, the steps for constructing a travel preference discrimination matrix based on passenger travel preferences and calculating passenger travel preference weights are as follows:

[0030] S31, calculate the relative importance of each pair of convenience preference, efficiency preference, and economy preference for the ticket-purchasing passenger i, and construct a preference discrimination matrix:

[0031] R = (r ij ) 3×3

[0032] In the formula, r ij This indicates the relative importance of travel preference i relative to travel preference j;

[0033] S32, Solve for the maximum eigenvalue λ of the preference discrimination matrix. imax and corresponding feature vectors Where j = 1, 2, 3;

[0034] S33, the feature vector After normalization, a weight vector W = {ω} of travel preferences is generated. 1i ,ω 2i ,ω 3i}, where ω 1i ,ω 2i ,ω 3i These represent the weights of passenger i's preferences for convenience, efficiency, and economy, respectively.

[0035] Furthermore, in step S4, the j-th travel plan attribute of the target route Y includes: the departure time S of the m-th journey. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm And calculate the transfer time G of the travel plan's modes of transportation. Yj The total travel time (H) of the travel plan Yj 1. Ticket price for the travel plan Yj ;

[0036] Among them, for the direct travel scheme in the multi-modal travel scheme, the transfer time G of the travel scheme is... Yj =0, the total travel time H of the trip plan Yj =S Yj1,2 -S Yj1,1 The ticket price for the travel plan IYj =Q Yj1 ;

[0037] For travel plans requiring transfers, first select a suitable set of transfer points, and then collect the departure time S of the m-th leg based on the set of transfer points. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm The transfer time, total travel time, and fare for the travel plan are calculated using the following formula:

[0038]

[0039] Furthermore, in step S5, the max-min method is used to standardize the multi-modal travel plan attributes of the ticket-purchasing passengers' target routes:

[0040]

[0041] In the formula, x = G, H, I, where G, H, and I represent the transfer time of the travel mode, the total travel time of the travel plan, and the ticket price of the travel plan, respectively; x j This represents the original value of a certain attribute of the j-th multimodal travel plan. Let x represent the standardized attribute value of the j-th multimodal travel plan. max x represents the maximum value of the attribute. min This represents the minimum value of the attribute.

[0042] Furthermore, in step S5, based on the standardized travel plan attributes, the optimal and worst solutions for the route travel plan are calculated:

[0043]

[0044] In the formula, j = 1, 2, ..., M, and M represents the number of travel options for the target route Y; This represents the maximum value of attribute x among all travel options for route Y. Let x represent the minimum value of attribute x among all travel options for route Y, where x = G, H, I;

[0045] Based on the travel preference weight vector W = {ω1, ω2, ω3}, calculate the Euclidean distance between the j-th route Y and the optimal and worst solutions:

[0046]

[0047] Further, in step S5, the score of the j-th scheme of route Y is calculated:

[0048]

[0049] Among them, S j The larger the value of S, the closer the solution j is to the optimal solution, and the higher it should be recommended in the order of preference; j The smaller the value, the closer the solution j is to the worst solution, and the lower the recommendation order. The solutions are sorted from high to low according to their scores to generate a personalized multi-modal travel solution sequence.

[0050] Compared with the prior art, the significant advantages of this invention are as follows:

[0051] 1. This invention identifies the travel characteristics of passengers based on their multi-modal historical travel records, quantitatively analyzes the travel preferences that influence passengers' ticket purchasing behavior, calculates the weights of passengers' travel preferences, and forms a personalized multi-modal travel plan sequence generation method.

[0052] 2. The personalized multi-modal travel plan sequence generator proposed in this invention will provide guidance and suggestions for improving the matching degree between personalized travel plans and passengers, thereby improving the efficiency of passengers' ticket purchase and having great significance for improving the service quality of online travel ticket purchase services. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the steps involved in implementing the present invention.

[0054] Figure 2 This is a schematic diagram of a multi-modal intercity travel scheme sequence generated in an embodiment of the present invention. Detailed Implementation

[0055] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0056] To address the issues of low personalization and limited travel modes in travel service platform recommendations when passengers purchase tickets online, this invention integrates feasible direct travel plans and connecting travel plans. Furthermore, it analyzes travel preference factors influencing passenger ticketing behavior, mining passengers' convenience, efficiency, and cost preferences based on their multi-modal historical travel records. Using the Analytic Hierarchy Process (AHP), weights are assigned to these preferences, ultimately achieving a more personalized multi-modal travel plan sequence generation method.

[0057] This invention fully considers the differences in travel preferences of each passenger. First, it identifies the travel characteristics of passengers based on their historical travel records. Then, it identifies individual passenger preferences and establishes a passenger preference discrimination matrix, calculating the weight allocation of individual preferences. Next, it collects attribute information on multi-modal travel plans based on the travel needs of ticket-buying passengers. Finally, it combines the passenger's travel preference weights and the attribute information of the multi-modal travel plans to score and rank the travel plans, generating a personalized travel plan sequence for the passenger. This invention helps improve the matching degree between personalized travel plans and passengers, enhances users' ticket purchasing efficiency, and thus achieves a win-win situation for both users and the platform, which is of great significance for improving the service quality of online travel ticketing services.

[0058] To achieve the above objectives, this invention provides a method for generating personalized travel plan sequences based on passenger travel characteristics, such as... Figure 1 As shown, it includes the following steps:

[0059] Step 1: Establishment of a passenger travel characteristic identification method

[0060] Collect the historical travel records of N passengers from the travel service platform, where passenger i has traveled N times. i Collect the route attribute features X of passenger i's j-th historical trip. ij (i = 1, 2, ..., N, j = 1, 2, ..., N) i This includes the transfer time TT for passenger i's j-th trip. ij The maximum historical transfer time TT for all passengers whose origin and destination are the same as passenger i's j-th trip. Aij If there is no transfer, skip this parameter; the total travel time IT for passenger i's j-th trip. ij IT is the maximum historical total travel time of all passengers whose origin and destination are the same as passenger i's j-th trip. Aij The ticket price P for passenger i's jth trip. ij The maximum historical travel fare P for all passengers whose journeys originate and terminate at the same point as passenger i's j-th trip. Aij .

[0061] Based on the travel route attribute characteristics of passenger i's j-th trip, a passenger travel characteristic identification method is established, including calculating the average transfer ratio A. ij Average time ratio B ij Average price ratio C ij :

[0062] A ij =TT ij / TT Aij (1)

[0063] B ij=IT ij / IT Aij (2)

[0064] C ij =P ij / P Aij (3)

[0065] The average transfer ratio, average time ratio, and average price ratio all range from [0,1].

[0066] The higher the average transfer ratio, the longer the transfer time for passenger i's historical travel plans, meaning the less acceptable the passenger is to longer transfer times. Similarly, the higher the average time ratio, the longer the total travel time for passenger i's historical travel plans, also meaning the less acceptable the passenger is to longer total travel time. Finally, the higher the average price ratio, the higher the price of passenger i's historical travel plans, meaning the less acceptable the passenger is to higher travel prices.

[0067] When a passenger i is purchasing a ticket on a travel service platform, it is determined whether passenger i has a historical travel record on the travel service platform, and the travel characteristics of passenger i are calculated accordingly.

[0068] True: If passenger i has N records in the travel service platform. i If there are 10 historical travel records, then calculate N. i The average travel characteristics of the previous trips are used as the travel characteristics of the ticket-purchasing passenger i:

[0069]

[0070] If the ticket purchaser i has no historical travel record on the travel service platform, then the average of all passenger travel characteristics is calculated as the travel characteristic of the ticket purchaser i.

[0071]

[0072] In the formula, A N B N C N These represent the average of the average transfer rate, average time rate, and average price rate for all passengers, respectively.

[0073] Step 2, passenger travel preference identification

[0074] Based on passenger i's travel characteristics, the travel preference factors influencing passenger i's ticket purchasing behavior are categorized into convenience preference (D). i Efficiency Preference E i Economic preference F i .

[0075] Based on the average transfer rate, average time rate, and average price rate from step 1, calculate the convenience preference characteristics, efficiency preference characteristics, and economy preference characteristics of the ticket-purchasing passenger i:

[0076]

[0077] When passenger i has a travel history on the travel service platform, the values ​​of A, B, and C are respectively: A i B i C i ;

[0078] When the ticket-purchasing passenger i has no historical travel record in the travel service platform, the values ​​of A, B, and C are respectively: A i ′、B i ′、C i ′;

[0079] Here, convenience preference represents passenger i's preference for the number of transfers and travel time in a travel plan, which differs from that of other passenger groups. The larger the value of the convenience preference index, the less time passenger i has spent on transfers in their historical travel records, meaning that passenger i is more sensitive to transfers and less willing to accept multiple transfers or long waiting times.

[0080] Efficiency preference represents passenger i's preference for the total travel time of a travel plan, which differs from that of the rest of the travel group. The larger the value of the efficiency preference index, the shorter the total travel time in passenger i's historical travel records, meaning that passenger i is more sensitive to the total travel time and less likely to accept travel plans that take too long.

[0081] Economic preference represents passenger i's preference for the total price of travel options, which differs from the overall passenger group. A higher value for economic preference indicates lower prices in passenger i's historical travel records, meaning passenger i is more price-sensitive and less likely to accept more expensive travel options.

[0082] Specifically, if ticket-purchasing passenger i has no historical travel records on the travel service platform, their travel characteristics are defined as the travel characteristics of the passenger group on the travel service platform. Therefore, the travel preference labeling result for this type of passenger is D. i =E i =F i =1.

[0083] Step 3: Calculate passenger travel preference weights

[0084] Using the Analytic Hierarchy Process (AHP), the relative importance of each pair of convenience preference, efficiency preference, and economy preference for ticket-purchasing passenger i is calculated, and a preference judgment matrix is ​​constructed:

[0085] R = (r ij ) 3×3 (13).

[0086] In the formula, r ij This indicates the relative importance of travel preference i compared to travel preference j.

[0087] Solving the formula based on the preference discrimination matrix:

[0088] RW=λW (14)

[0089] Where λ represents the eigenvalue and W represents the corresponding eigenvector.

[0090] Calculate the maximum eigenvalue λ of the preference discrimination matrix. imax The corresponding feature vectors are then normalized to generate a weight vector for travel preferences.

[0091] W={ω 1i ,ω 2i ,ω 3i} (15)

[0092] Where, ω 1i ω 2i ω 3i These represent the weights of passenger i's convenience preference, efficiency preference, and economy preference, respectively.

[0093] Step 4: Collection of ticketing information for multi-modal travel plans

[0094] When a passenger i is purchasing a ticket on a travel service platform, the ticketing attribute information of the travel plans is collected based on the M multi-mode travel plans of the passenger i's desired route. The travel plans include travel modes such as direct train, direct bus, train-to-train transfer, and train-to-bus transfer.

[0095] The j-th travel option attribute of the target route Y includes the departure time S of the m-th segment. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm And calculate the transfer time G of the travel plan's modes of transportation. Yj The total travel time (H) of the travel plan Yj 1. Ticket price for the travel plan Yj .

[0096] Among them, for direct travel options in multimodal travel plans, the transfer time G is... Yj =0, the total travel time H of the trip plan Yj =S Yj1,2 -S Yj1,1 The ticket price for the travel plan IYj =Q Yj1 .

[0097] For travel plans requiring transfers, first select a suitable set of transfer points, and then collect the departure time S of the m-th leg based on the set of transfer points. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm The transfer time, total travel time, and fare for the travel plan are calculated using the following formula:

[0098]

[0099] Step 5: Generation of personalized multimodal travel plan sequences

[0100] The ticketing attribute information of the multi-modal travel plan in step 4 is standardized:

[0101] The max-min method is used to standardize the multimodal travel scheme attributes of ticket-purchasing passengers' target routes in step 4.

[0102]

[0103] In the formula, G, H, and I represent the transfer time of the travel mode, the total travel time of the travel plan, and the ticket price of the travel plan, respectively; x j This represents the original value of a certain attribute of the j-th multimodal travel plan. Let x represent the standardized attribute value of the j-th multimodal travel plan. max x represents the maximum value of the attribute. min This represents the minimum value of the attribute.

[0104] Calculate the travel plan evaluation index for the route:

[0105] Based on the standardized travel plan attributes, the optimal solution Z of the route travel plan is calculated. + And the worst solution Z - :

[0106]

[0107] In the formula, M represents the number of travel options for the target route Y. This represents the maximum value of attribute x among all travel options for route Y. This represents the minimum value of attribute x among all travel options for route Y.

[0108] Combining the travel preference weight vector W = {ω1, ω2, ω3} from step 3, calculate the Euclidean distance between the j-th option of the target route Y and the optimal and worst solutions:

[0109]

[0110] Where, d j + Let d be the Euclidean distance between the j-th route to the target route Y and the optimal solution. j - The Euclidean distance between the j-th route to the target route Y and the worst solution.

[0111] The TOPSIS method is used to generate a sequence of personalized travel plans. The process is as follows:

[0112] Calculate the score S of the j-th option of route Y. j :

[0113]

[0114] Among them, S j The larger the value of S, the closer the solution j is to the optimal solution, and the earlier it should be recommended. j A smaller value indicates that solution j is closer to the worst solution, and its recommendation order is lower. The solutions are then sorted from highest to lowest score to generate a personalized multimodal travel solution sequence.

[0115] This embodiment takes a passenger, O, who has a travel need from Nanjing, Jiangsu Province to Huai'an, Jiangsu Province on March 11, 2023 as an example. The passenger purchases a ticket through an online ticketing app, and has a previous ticket purchase record for the same route on that app. The implementation steps are as follows:

[0116] B1, Establishment of Passenger Travel Feature Identification Method

[0117] First, based on passenger o's travel needs and route, collect N historical order records from the online ticketing app for the journey from Nanjing, Jiangsu Province to Huai'an, Jiangsu Province. The transfer time for passenger i's j-th trip is TT. ij The maximum transfer time for all passengers traveling to the same origin and destination as passenger i's j-th trip is TT. Aij If there is no transfer, skip this parameter; the total travel time for passenger i's j-th trip is IT. ij The maximum total travel time for all passengers whose journeys originate and terminate at the same point as passenger i's j-th trip is IT. Aij The ticket price for passenger i's j-th trip is P. ij The maximum value of the ticket prices for all passengers traveling to the same origin and destination as passenger i on their j-th trip is P. Aij A summary of historical orders is shown in Table 1.

[0118] Table 1 Example of Historical Order Summary

[0119]

[0120] Based on passenger o's historical travel records from Nanjing to Huai'an in the online ticketing app, and according to the historical transfer times and the maximum historical transfer times for the same origin and destination in Table 1, the average transfer ratio of passenger o's historical trips is calculated as follows:

[0121]

[0122] Based on the historical total travel time and the maximum historical total travel time for the same origin and destination in Table 1, calculate the average time ratio B of passenger o's historical trips. ij ,have:

[0123]

[0124] Based on the historical travel fares in Table 1 and the maximum historical travel fare for the same origin and destination, calculate the average price ratio C of passenger o's historical travels. ij ,have:

[0125]

[0126] Then, it is determined whether the ticket-purchasing passenger has a historical travel record on the travel service platform. Since the ticket-purchasing passenger o mentioned in the assumption has a previous ticket purchase record for the same route on this ticket-purchasing APP, the travel characteristics of passenger o from Nanjing to Huai'an are calculated as follows:

[0127]

[0128] B2, Identification of Passenger Travel Preferences

[0129] Based on the behavioral characteristics of ticket-purchasing passenger o in step B1, their travel preference characteristics are identified, including convenience preference D. i Efficiency Preference E i Economic preference F i .

[0130] First, calculate the travel characteristics of all passengers, that is, calculate the average of the average transfer ratio, average time ratio, and average price ratio for all passengers:

[0131]

[0132] Then, based on the travel characteristics of the ticket-purchasing passenger o, the travel preferences of the ticket-purchasing passenger are identified as follows:

[0133]

[0134] B3, Calculate the weights of ticket-buying passengers' travel preferences.

[0135] Based on the ticket purchase preference characteristics of the ticket-buying passenger o calculated in step B2, the relative importance of each pair of convenience preference, efficiency preference, and economy preference is calculated using the analytic hierarchy process (AHP). The resulting travel preference discrimination matrix R = (r... ij ) 3×3 The relative importance of convenience preference to efficiency preference can be calculated by the following formula:

[0136] r 12 =D o / E o =0.532 / 0.601=0.885.

[0137] Table 2 Travel Preference Judgment Matrix

[0138] <![CDATA[R=(r ij ) 3×3 ]]> Convenience preference Efficiency Preference Economic preferences Convenience preference 1 0.885 0.580 Efficiency Preference 1.130 1 0.655 Economic preferences 1.724 1.526 1

[0139] Solve for the eigenvalues ​​and eigenvectors of the preference discrimination matrix, i.e., solve for RW = λW, where λ represents the eigenvalue and W represents the corresponding eigenvector. The largest eigenvalue is λ. max =3, the corresponding eigenvector is W max ={0.801,0.905,1.380}, normalized using the following formula:

[0140]

[0141] The preference weight vector W = {0.260, 0.293, 0.447} for passenger o is obtained, where the three values ​​represent the preference weights for convenience, efficiency, and economy, respectively. It can be seen that passenger o has a higher requirement for economy than for efficiency, and a higher requirement for efficiency than for convenience.

[0142] B4, Multimodal travel solution ticketing information collection

[0143] Based on the needs of the ticket-purchasing passenger o, ticket attribute information Y for M multi-modal travel routes from Nanjing to Huai'an is collected. j j = 1, 2, ..., M, including direct train, direct bus, train-to-train transfer, and train-to-bus transfer modes of transportation.

[0144] For direct travel plans, calculate the total travel time H. Yj The ticket price for the travel plan I Yj For travel plans requiring transfers, first select a suitable set of transfer locations, and then calculate the transfer time G for the travel plan. Yj Total time H Yj Ticket Price I YjTable 3 shows the attribute information of some multimodal travel options.

[0145] Table 3. Example of attribute information for multimodal travel plans

[0146]

[0147] B5, Personalized Generation of Multimodal Travel Plans

[0148] Based on the data in Table 3, the j-th multimodal travel scheme is standardized using the max-min method:

[0149]

[0150] In the formula, x j This represents the original value of a certain attribute of the j-th multimodal travel plan. Let x represent the standardized attribute value of the j-th multimodal travel plan. max x represents the maximum value of the attribute. min This represents the minimum value of the attribute. Table 4 shows some of the standardized results.

[0151] Then, based on the passenger's ticket purchase demand route Y and multi-modal travel plan ticketing information described in step B4, the optimal solution Z of the multi-modal travel plan for the corresponding route is calculated. + And the worst solution Z - :

[0152]

[0153] In the formula, This represents the maximum value of attribute x among all multimodal travel options for route Y. Let x be the minimum value of attribute x among all multimodal travel options for route Y. Since the data is standardized, the optimal solution is [1,1,1] and the worst solution is [0,0,0].

[0154] Combined with the weight vector W = {ω} described in step B3 1o ,ω 2o ,ω 3o} Calculate the Euclidean distance between the j-th route Y and the optimal and worst solutions:

[0155]

[0156] In the formula, x = TT Y IT Y ,P Y respectively with ω 1o ,ω 2o ,ω 3o correspond.

[0157] Furthermore, based on the Euclidean distances between the j-th solution and the optimal and worst solutions of route Y, the multimodal travel schemes of route Y are scored. The score for the j-th solution is as follows:

[0158]

[0159] Table 4 shows the final calculated scores, the standardized multimodal travel scheme attributes, and the preference weights of passenger o.

[0160] Table 4. Example of scoring results for multimodal travel plans

[0161]

[0162] Among them, S j The larger the value of S, the closer the solution j is to the optimal solution, and the earlier it should be recommended. j A smaller value indicates that solution j is closer to the worst solution, and therefore should be recommended later. The score S for multimodal travel solutions... j The travel options are sorted from highest to lowest to generate a personalized multimodal travel plan recommendation sequence for ticket-buying passengers. Based on the five travel options described in Table 4, the generated recommendation sequence is as follows: Figure 2 As shown.

[0163] Table 4 shows that even though option 4 is a connecting flight, it receives a higher score based on passenger preference weighting, thus making it a priority recommendation. Current online ticketing apps prioritize direct travel options over passenger preferences when recommending travel plans. In contrast, the method of this invention improves the personalized matching of travel plan recommendations.

Claims

1. A method for generating personalized travel plan sequences based on passenger travel characteristics, characterized in that, Includes the following steps: S1, collect historical travel records of N passengers in the travel service platform, wherein the travel times of passenger i is N i ; then determine whether the ticketed passenger i has historical travel records in the travel service platform, and calculate the travel characteristics of passenger i according to the travel characteristic recognition method; S2, combining passenger travel characteristics, divides the travel preference factors influencing passenger ticketing behavior into convenience preference D. i Efficiency Preference E i and economic preference F i Then, based on the travel characteristics of passenger i, the travel preferences of passenger i are identified. S3. Using the analytic hierarchy process (AHP), the relative importance of each pair of travel preferences is calculated to construct a preference discrimination matrix. The weight vector W = {ω1, ω2, ω3} of travel preferences is calculated, where ω1, ω2, and ω3 are the weights of convenience preference, efficiency preference, and economy preference, respectively. S4, based on the ticket-purchasing passenger's target route Y, obtain multi-mode travel plans for the route; wherein, the ticketing attribute information of the j-th travel plan for the target route Y includes: the total travel time H of the travel plan. Yj Ticket Price I Yj and the transfer time G of the transfer scheme Yj ; S5, standardize the ticketing attribute information of the multi-modal travel plan, and then calculate the optimal solution Z of the target route Y travel plan. + And the worst solution Z - And sequentially calculate the outgoing travel options for route Y and the optimal solution Z. + And the worst solution Z - The Euclidean distance between them is calculated; then, the TOPSIS method is used, combined with the travel preference weight vector W={ω1,ω2,ω3}, to rank the multimodal travel schemes and generate a personalized travel scheme sequence for ticket-purchasing passengers; the optimal and worst solutions of the target route Y travel scheme are calculated as follows: In the formula, j = 1, 2, ..., M, and M represents the number of travel options for the target route Y; This represents the maximum value of attribute x among all travel options for route Y. Let x represent the minimum value of attribute x among all travel options for route Y, where x = G, H, I; G, H, and I represent the transfer time of the travel mode, the total travel time of the travel option, and the ticket price of the travel option, respectively. Based on the travel preference weight vector W = {ω1, ω2, ω3}, calculate the Euclidean distance between the j-th route of the target route Y and the optimal and worst solutions: The score for the j-th option of the target route Y is calculated as follows: Among them, S j The larger the value of S, the closer the solution j is to the optimal solution, and the higher it should be recommended in the order of preference; j The smaller the value, the closer the solution j is to the worst solution, and the lower the recommendation order. The solutions are sorted from high to low according to their scores to generate a personalized multi-modal travel solution sequence. d represents the standardized attribute value of the j-th multimodal travel plan; j + Let d represent the Euclidean distance between the j-th solution and the optimal solution. j - Let represent the Euclidean distance between the j-th solution and the worst solution.

2. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 1, characterized in that, In step S1, historical travel records of N passengers are collected, and the travel characteristics of passenger i's j-th historical trip are calculated; the travel characteristics include: average transfer ratio A. ij Average time ratio B ij and average price ratio C ij Their expressions are as follows: A ij = TT ij / T Aij B ij = IT ij / IT Aij C ij = P ij / P Aij In the formula, TT ij TT represents the transfer time for passenger i's j-th historical trip; Aij This represents the maximum historical transfer time for all passengers traveling to the same origin and destination. If there is no transfer, this parameter is skipped. ij IT represents the total travel time of passenger i's j-th historical trip. Aij P represents the maximum historical total travel time for all passengers originating from and destined for the same destination; ij P represents the ticket price for passenger i's j-th historical trip. Aij This represents the maximum historical travel fare for all passengers traveling to the same origin and destination.

3. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 1, characterized in that, In step S1, it is determined whether the ticket-purchasing passenger i has a historical travel record in the travel service platform. The determination principle is as follows: If the judgment result is true, and passenger i has N i If there are 10 historical travel records, then calculate N. i The average travel characteristics of each historical trip are used as the travel characteristics of the ticket-purchasing passenger i, and their expressions are as follows: If the judgment result is false, the average value of all passenger travel characteristics is calculated as the travel characteristic of the ticket-purchasing passenger i, and the expressions are as follows: In the formula, A N , B N , C N respectively represent the average values of the average transfer ratio, the average time ratio, and the average price ratio of all passengers.

4. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 3, characterized in that, In step S2, the process of identifying passenger travel preferences based on the travel characteristics of the ticket-purchasing passengers is as follows: S21, according to the travel characteristics of passenger i, the travel preference factors affecting the ticket purchase behavior of the passenger i are divided into convenience preference D i , efficiency preference E i and economic preference F i , whose expressions are as follows: When the ticket passenger i has a historical travel record in the travel service platform, the values of A, B and C are respectively: i , B i , C i ; When the ticket passenger i has no historical travel record in the travel service platform, the values of A, B and C are respectively: i A i B i C.

5. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 1, characterized in that, In step S3, the steps for constructing a travel preference discrimination matrix and calculating the travel preference weights based on passenger travel preferences are as follows: S31, calculate the relative importance of each pair of convenience preference, efficiency preference, and economy preference for the ticket-purchasing passenger i, and construct a preference discrimination matrix: R＝(r ij ) 3×3 wherein r ij denotes the relative importance of travel preference i with respect to travel preference j; S32, Solve for the maximum eigenvalue λ of the preference discrimination matrix. imax and corresponding feature vectors Where j = 1, 2, 3; S33, the feature vector After normalization, a weight vector W = {ω} of travel preferences is generated. 1i ,ω 2i ,ω 3i }, where ω 1i ,ω 2i ,ω 3i These represent the weights of passenger i's preferences for convenience, efficiency, and economy, respectively.

6. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 1, characterized in that, In step S4, the j-th travel plan attribute of the target route Y includes: the departure time S of the m-th journey. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm And calculate the transfer time G of the travel plan's modes of transportation. Yj The total travel time (H) of the travel plan Yj 1. Ticket price for the travel plan Yj ; Among them, for the direct travel scheme in the multi-modal travel scheme, the transfer time G of the travel scheme is... Yj =0, the total travel time H of the trip plan Yj =S Yj1,2 -S Yj1,1 The ticket price for the travel plan I Yj =Q Yj1 ; For travel plans requiring transfers, first select a suitable set of transfer locations, and then collect the departure time S of the m-th leg based on the set of transfer locations. Yjm,1 and arrival time S Yjm,2 The ticket price Q for the m-th leg Yjm The transfer time, total travel time, and fare for the travel plan are calculated using the following formula:

7. The method for generating personalized travel plan sequences based on passenger travel characteristics according to claim 1, characterized in that, In step S5, the max-min method is used to standardize the multi-modal travel plan attributes of the ticket-purchasing passengers' target routes: In the formula, x j Let x represent the original value of a certain attribute of the j-th multimodal travel plan. max x represents the maximum value of the attribute. min This represents the minimum value of the attribute.

Images