Intermodal scheme optimization method, training method, device and electronic equipment

Abstract

The embodiment of the present application provides an intermodal scheme optimization method, training method, device and electronic device, which relates to the field of transportation technology. The method can optimize and filter the intermodal schemes that are initially recommended to ensure the rationality of the schemes. The method can specifically obtain the normalized second transfer point location information based on the first departure point location information, the first arrival point location information and the first transfer point location information included in the intermodal scheme to be optimized; input the second transfer point location information into a trained classification model; obtain the predicted value of the second transfer point location information for multiple intermodal modes through the classification model; obtain the initial score of the intermodal scheme to be optimized; calculate the evaluation score of the intermodal scheme to be optimized based on the initial score and the predicted value; re-sort the intermodal schemes to be optimized according to the evaluation score, and filter out the intermodal schemes with evaluation scores below the threshold.

Description

Technical Field

[0001] The present application relates to the field of transportation technology, and in particular to a method for optimizing an intermodal transport solution, a training method, a device and an electronic device. Background Art

[0002] The intermodal transport of passengers by rail and other modes of transportation can bring more convenience to passengers and improve the user experience. For example, air-rail intermodal transport is the combination of air transport and rail transport, providing passengers with more efficient and diversified travel options.

[0003] At present, artificial intelligence can be used to recommend a variety of candidate intermodal transport solutions to passengers. The rationality of the intermodal transport solutions recommended by the model needs to be evaluated. Due to various reasons, some of the current evaluation solutions may have poor or unreasonable queue sorting of the overall recommended solutions, or the recommended transit cities may not be suitable, resulting in unreliable recommended solutions. They need to be optimized to improve the quality of the recommended intermodal transport solutions. Summary of the invention

[0004] In view of this, the embodiments of the present application provide a method, a training method, a device and an electronic device for optimizing an intermodal transport scheme to solve the above-mentioned problems, optimize and filter the initially recommended intermodal transport scheme, and ensure the rationality and reliability of the intermodal transport scheme.

[0005] In a first aspect, an embodiment of the present application provides a method for optimizing an intermodal transport scheme, which is applied to an electronic device, and comprises: obtaining normalized second transfer point location information according to the first departure point location information, the first arrival point location information and the first transfer point location information included in the intermodal transport scheme to be optimized; the intermodal transport scheme to be optimized also includes an intermodal transport mode; wherein the intermodal transport mode is one of a plurality of intermodal transport modes, and the plurality of intermodal transport modes include passengers first taking a first means of transportation and then transferring to a second means of transportation, or passengers first taking the second means of transportation and then transferring to the first means of transportation; the first means of transportation and the second means of transportation are means of transportation based on different modes of transportation; the second transfer point location information is input into a trained classification model; through the classification model, the predicted values ​​of the second transfer point location information for a plurality of intermodal transport modes are obtained; an initial score of the intermodal transport scheme to be optimized is obtained; based on the initial score and the predicted value, an evaluation score of the intermodal transport scheme to be optimized is calculated; according to the evaluation score, the intermodal transport schemes to be optimized are reordered, and intermodal transport schemes with evaluation scores lower than a threshold are filtered out.

[0006] The above method can optimize the sorting of multiple candidate intermodal transport solutions, filter out unreasonable intermodal transport solutions, improve the reliability and quality of the intermodal transport solutions recommended to users, and enhance user experience.

[0007] In a second aspect, an embodiment of the present application also provides a classification model training method, which is applied to an electronic device, and the method includes: obtaining multiple intermodal transport schemes as sample data; each of the multiple intermodal transport schemes includes a first departure point location information, a first arrival point location information, a first transfer point location information and an intermodal transport mode; according to the first departure point location information, the first arrival point location information, and the first transfer point location information, obtain normalized second transfer point location information; cluster the multiple second transfer point location information to obtain a first cluster and a second cluster; mark the transfer points in the first cluster with a label of the first intermodal transport mode, and mark the transfer points in the second cluster with a label of the second intermodal transport mode; mark other transfer points that are not clustered into the first cluster or the second cluster with other labels; wherein the first intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation; the second intermodal transport mode is that the passenger first takes the second means of transportation and then transfers to the first means of transportation; use the labeled second transfer point location information as a training sample to train the classification model to obtain a trained classification model.

[0008] In one possible implementation, according to the first departure point location information, the first arrival point location information, and the first transfer point location information, the normalized second transfer point location information is obtained, including: according to the first departure point location information, the first arrival point location information, and the first transfer point location information, calculating the first distance, the second distance, and the third distance; wherein the first distance is the distance between the first departure point and the first transfer point, the second distance is the distance between the first transfer point and the first arrival point; the third distance is the distance between the first departure point and the first arrival point; calculating any of the first distance, the second distance, and the third distance When the sum of two distances is greater than the other distance, determine that the first starting point, the first arrival point, and the first transfer point form a triangle; map the first starting point and the first arrival point to two points specified in the coordinate system; the distance between the first starting point and the first arrival point is K times the distance between the two specified points; reduce the first distance and the second distance by K times respectively to obtain the first side length and the second side length; calculate the coordinates of the transfer point in the coordinate system according to the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the obtained coordinates of the transfer point are used as the normalized second transfer point information.

[0009] In one possible implementation, after calculating the first distance, the second distance, and the third distance, and before determining that the first starting point, the first arrival point, and the first transfer point form a triangle, the method further includes: determining that the first starting point, the first arrival point, and the first transfer point do not form a triangle when the sum of any two distances among the first distance, the second distance, and the third distance is less than or equal to the other distance; and increasing the shortest distance among the first distance, the second distance, and the third distance by a preset length.

[0010] In one possible implementation, clustering the plurality of second transfer point location information includes: clustering the plurality of second transfer point location information using a density-based application clustering algorithm DBSCAN.

[0011] In a third aspect, an embodiment of the present application further proposes a device for optimizing an intermodal transport scheme, the device comprising: a normalization unit, for obtaining normalized second transfer point position information according to the first departure point position information, the first arrival point position information and the first transfer point position information included in the intermodal transport scheme to be optimized; the intermodal transport scheme to be optimized also comprises an intermodal transport mode; wherein the intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation, or the passenger first takes the second means of transportation and then transfers to the first means of transportation; the first means of transportation and the second means of transportation are means of transportation based on different modes of transportation; a prediction unit, for inputting the second transfer point position information into a trained classification model; through the classification model, obtaining a prediction value of the second transfer point position information relative to the category corresponding to the first intermodal transport mode; an acquisition unit, for obtaining an initial score of the intermodal transport scheme to be optimized; an evaluation unit, for calculating an evaluation score of the intermodal transport scheme to be optimized based on the initial score and the prediction value; and, according to the evaluation score, reordering the intermodal transport schemes to be optimized, and filtering out the intermodal transport schemes with evaluation scores lower than a threshold.

[0012] In a fourth aspect, an embodiment of the present application further proposes a classification model training device, which includes: an acquisition unit, used to acquire multiple intermodal transport schemes as sample data; each of the multiple intermodal transport schemes includes a first departure point location information, a first arrival point location information, a first transfer point location information and a first intermodal transport mode; a normalization unit, used to obtain normalized second transfer point location information based on the first departure point location information, the first arrival point location information and the first transfer point location information; a clustering unit, used to cluster multiple second transfer point location information to obtain a first cluster and a second cluster; the transfer points in the first cluster are marked with a label of the first intermodal transport mode, and the transfer points in the second cluster are marked with a label of the second intermodal transport mode; other transfer points that are not clustered into the first cluster or the second cluster are marked with labels of other intermodal transport modes; wherein the first intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation; the second intermodal transport mode is that the passenger first takes the second means of transportation and then transfers to the first means of transportation; a training unit, used to use the labeled second transfer point location information as a training sample to train the classification model to obtain a trained classification model.

[0013] In one possible implementation, the normalization unit is specifically used to: calculate the first distance, the second distance and the third distance according to the first starting point location information, the first arrival point location information and the first transfer point location information; wherein the first distance is the distance between the first starting point and the first transfer point, the second distance is the distance between the first transfer point and the first arrival point; the third distance is the distance between the first starting point and the first arrival point; when the sum of any two distances among the first distance, the second distance and the third distance is greater than the other distance, determine that the first starting point, the first arrival point and the first transfer point form a triangle; map the first starting point and the first arrival point to two points specified in the coordinate system; the distance between the first starting point and the first arrival point is K times the distance between the two specified points; reduce the first distance and the second distance by K times respectively to obtain the first side length and the second side length; calculate the coordinates of the transfer point in the coordinate system according to the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the coordinates of the transfer point obtained are used as the normalized second transfer point information.

[0014] In one possible implementation, after calculating the first distance, the second distance, and the third distance, and before determining that the first starting point, the first arrival point, and the first transit point form a triangle, the normalization unit is further used to: determine that the first starting point, the first arrival point, and the first transit point do not form a triangle when the sum of any two distances among the first distance, the second distance, and the third distance is less than or equal to the other distance; and increase the shortest distance among the first distance, the second distance, and the third distance by a preset unit length.

[0015] In one possible implementation, when clustering the plurality of second transfer point location information, the clustering unit is specifically used to: cluster the plurality of second transfer point location information using a density-based application clustering algorithm DBSCAN.

[0016] In a fifth aspect, an embodiment of the present application further provides an electronic device, the electronic device comprising: a processor, the processor being used to execute a computer program or instruction in a memory to implement a method as described in any one of the first or second aspects above.

[0017] In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed by a processor, the method as described in any one of the first aspect or the second aspect above is implemented.

[0018] In the seventh aspect, an embodiment of the present application further provides a computer program product, which includes a program. When the program is executed by an electronic device, the electronic device implements the method as described in any one of the first aspect or the second aspect above.

[0019] In the eighth aspect, an embodiment of the present application also provides a chip system, comprising: a communication interface for inputting and / or outputting data; a processor for executing a computer executable program so that a device equipped with the chip system executes a method as described in any one of the first or second aspects above. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0021] Figure 1 Schematic diagrams of the flow charts corresponding to the training phase and the prediction phase in some embodiments of the method provided in the embodiments of the present application;

[0022] Figure 2 A schematic diagram of a normalized process flow in some embodiments of the method provided in the embodiments of the present application;

[0023] Figure 3 A schematic diagram of the normalized distribution of transfer point positions in some embodiments of the training method provided in the embodiments of the present application;

[0024] Figure 4 A schematic diagram of two clusters after normalized transfer point position clustering in some embodiments of the training method provided in the embodiments of the present application;

[0025] Figure 5 A schematic diagram of some embodiments of the combined transport scheme optimization method provided in the embodiments of the present application and the application of the recommendation module in combination;

[0026] Figure 6 A schematic diagram of prediction results of a classification model in some embodiments of the method for optimizing the intermodal transport scheme provided by the embodiments of the present application;

[0027] Figure 7 Schematic diagram of several intermodal transport schemes recommended in some embodiments of the intermodal transport scheme optimization method provided in the embodiments of the present application;

[0028] Figure 8 A flow chart of a method for optimizing a combined transport solution provided in an embodiment of the present application;

[0029] Fig. 9 A flowchart of a model training method provided in an embodiment of the present application;

[0030] Fig.10 A flowchart of another feasible way of normalizing in some embodiments of the method provided in the embodiments of the present application;

[0031] Fig.11 A schematic diagram of the software module architecture of the intermodal transport solution optimization device provided in an embodiment of the present application;

[0032] Fig.12 A schematic diagram of the software module architecture of the classification model training device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to better understand the technical solution of the present application, the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0034] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0035] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0036] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0037] Before introducing the embodiments of the present application, the related technologies and their technical problems are first explained.

[0038] Intermodal transport refers to the combination of two or more modes of transport. Among them, intermodal transport of passengers by rail and other modes of transport is a business type with a large demand. In particular, air-rail intermodal transport, as an innovative travel mode, provides passengers with a more convenient and efficient travel plan. At present, there are more and more studies and practices on air-rail intermodal transport, but there are many deficiencies in these studies and practices.

[0039] For example, in the process of recommending the overall air-rail transport plan, the historical air-rail transport travel data of passengers will be used to train the machine learning model. Due to the preference of passengers, data imbalance will occur, causing the model to learn one-sided knowledge. The trained model will be more inclined to predict a specific part of the data, resulting in other more reasonable travel plans being at the back of the plan queue. On the other hand, evaluating the performance of the model is also a key step to improve the performance of the model. At present, the evaluation functions used by the model, such as accuracy, precision, recall, F1 score, etc., can generally only evaluate the first predicted plan, and cannot evaluate the second and subsequent plans predicted by the model.

[0040] Under the influence of data imbalance and limitations of model evaluation functions, it is easy to cause poor queue sorting of multiple recommended candidate intermodal solutions or the emergence of unreasonable solutions.

[0041] In addition, if the air-rail transport network is traversed, there may be many feasible transit cities. If the next step is performed for each transit city, the amount of calculation will increase significantly, which will require more computing resources and bring certain delays, and the time cost and computing cost are both high. Although some current traditional methods will set fixed distance or time parameters, and feasible transit cities beyond this parameter range will not be operated in the next step, thereby reducing the amount of calculation, but the setting of this parameter range is generally based on human experience, which is subjective and one-sided. Inappropriate transit cities are often recommended, while appropriate transit cities are not recommended. The quality of the recommended solutions is not high, and the reliability needs to be improved.

[0042] It can be seen that an optimization method is needed to optimize the recommended multiple intermodal transport schemes to improve the reliability and quality of the recommended schemes.

[0043] In view of this, an embodiment of the present application designs a method for optimizing intermodal transport schemes. The method is an optimization and filtering algorithm for air-rail intermodal transport recommendation schemes based on spatial features. The transfer points actually selected by passengers (such as transfer cities) are used as sample data, and the normalization method and cluster analysis method are used to obtain the classification and labeling of the transfer points in the sample data. The labeled sample data is used as training data to train the classification model, and a classification model can be obtained that predicts the category of the intermodal transport mode based on the transfer point location information in the recommended intermodal transport scheme. Combining the predicted value of the classification model with the initial score of the intermodal transport scheme, a comprehensive evaluation score is obtained. According to the comprehensive evaluation score, the recommended multiple intermodal transport schemes are re-sorted, and unreasonable intermodal transport schemes are filtered out, thereby realizing the optimization and filtering of the air-rail intermodal transport recommendation schemes.

[0044] It should be noted that the method proposed in the embodiment of the present application can be applied to intermodal transport scenarios of various combinations of different transport modes, for example, it can be applied to intermodal transport of railways and aviation, or intermodal transport of railways and highways, or intermodal transport of railways and ocean freight, or intermodal transport of aviation and ocean freight, etc. For ease of description, the following description will be mainly based on the intermodal transport scenario of railways as an example.

[0045] The intermodal transport scheme to be optimized at least includes information such as departure point information, arrival point information, transfer point information, and intermodal transport mode.

[0046] Among them, the intermodal mode refers to the combination of multiple modes of transportation corresponding to multiple trips of the same user. In the case of only one transfer point, the intermodal mode includes the passenger first taking the first means of transportation from the departure point to the transfer point, and then transferring to the second means of transportation from the transfer point to the arrival point; or the passenger first taking the second means of transportation from the departure point to the transfer point, and then transferring to the first means of transportation from the transfer point to the arrival point. The first means of transportation and the second means of transportation are transportation based on different modes of transportation.

[0047] For example, in the scenario of air and rail intermodal transport, when there is only one transfer point between the departure point and the arrival point, there are two intermodal transport modes: rail first, then air (rail-air), or air first, then rail (air-rail). That is, taking a train or high-speed rail for the first leg of the journey and a plane for the second leg is one intermodal transport mode; taking a plane for the first leg of the journey and a train or high-speed rail for the second leg is another intermodal transport mode. In the scenario of rail-air intermodal transport, the intermodal transport mode can be rail-air or air-rail.

[0048] The departure point information includes the departure point name and the departure point location information, the arrival point information includes the arrival point name and the arrival point location information, and the transfer point information includes the transfer point name and the transfer point location information.

[0049] The names of the departure point and the arrival point are generally determined according to user operations. For example, the departure point and the arrival point may be received through an interactive interface, which is input or selected by the user.

[0050] It should be noted that in some embodiments of the present application, the departure point, arrival point and transfer point are generally cities, for example, cities at the county level or above. In other embodiments, the arrival point, departure point or transfer point may also be a railway station, airport, transportation hub, high-speed rail station or other site.

[0051] To avoid confusion, Figure 1 As shown, the classification model training method and the intermodal transport scheme optimization method proposed in the embodiments of the present application are explained from the training stage and the application stage respectively.

[0052] Training phase:

[0053] like Figure 1The processing flow corresponding to the training stage shown in the figure, the classification model training method proposed in the embodiment of the present application can first obtain sample data of multiple intermodal transport schemes from the historical actual travel data of passengers, and then perform normalization processing based on the intermodal transport scheme sample data to obtain normalized transfer point location information, and cluster the multiple normalized transfer point location information to obtain the intermodal transport mode labels corresponding to the multiple transfer points, and then use the labeled transfer point location information as training samples to train the classification model to obtain a trained classification model.

[0054] like Figure 1 The processing flow corresponding to the prediction stage shown in the figure, the intermodal scheme optimization method proposed in the embodiment of the present application, for the intermodal scheme to be optimized, first normalizes it to obtain the normalized transfer point location information, inputs the transfer point location information into the classification model, and outputs the prediction value corresponding to the transfer point location information through the classification model. Then, based on the prediction value and the initial score of the intermodal mode to be optimized, a comprehensive evaluation score is calculated, and according to the evaluation score, multiple intermodal schemes to be optimized are re-ranked, and schemes with scores below the threshold are filtered out to complete the scheme optimization.

[0055] A specific embodiment is listed below.

[0056] S101, obtaining a plurality of intermodal transport solution sample data.

[0057] In this embodiment, multi-source heterogeneous data is first obtained, and intermodal transport solution sample data is obtained from the multi-source heterogeneous data. Multi-source heterogeneous data includes system intermodal transport data (or product intermodal transport data) and autonomous intermodal transport data. System intermodal transport data includes multiple intermodal transport orders, for example, multiple intermodal transport orders generated in the air-rail intermodal transport system. Autonomous intermodal transport data includes multiple groups of orders, one group of orders includes two non-intermodal transport orders corresponding to the same user, the time difference between the two non-intermodal transport orders is less than the predetermined time, and the arrival point of one of the two non-intermodal transport orders is in the same city as the departure point of the other order.

[0058] For example, in this embodiment, on the one hand, actual system intermodal transport data for a period of time is directly obtained from the system providing intermodal transport services, such as obtaining system intermodal transport data (or product intermodal transport data) from October 2023 to December 2023 from the air-rail intermodal transport service system.

[0059] On the other hand, we obtained order data from October 2023 to December 2023 from the train ticket sales system and the air ticket sales system, and combined the data of the same passenger's two trips with a time difference less than the scheduled duration and the arrival point of the previous trip and the departure point of the next trip to form autonomous intermodal data according to the passenger travel dimension. A total of 444,562 product intermodal data and autonomous intermodal data were obtained.

[0060] For example, using the user's account name or identity ID as the index, match any two trips with the same index on the same date to see if they meet the following conditions. If they do, merge them into one set of data:

[0061] The arrival point of the first trip of a user (or passenger) coincides with or is the same as the departure point of the second trip; the arrival point and departure point information can be obtained from the orders corresponding to the two trips;

[0062] The time difference between the actual arrival time of the first segment of the journey and the actual departure time of the second segment of the journey is less than the predetermined duration, which may be 2.5-5.5 hours, for example, 4 hours.

[0063] If the above two conditions are met, the two orders are combined into a set of data. This part of the data is based on the actual situation that the user has a transfer during the trip, but the user has not placed an order through the intermodal system. This part of the data can supplement the data source and make up for the defects of insufficient or single data source.

[0064] Based on the structure of product intermodal transport data, the autonomous intermodal transport data is processed into the same data format as the product intermodal transport data to obtain data in a unified format, in which each data item consists of train ticket order information and flight ticket order information.

[0065] For example, a train ticket order information may include one or more of the following items: travel date, departure train number, travel city, travel station, travel station telegraph code, departure date, arrival city, arrival station, arrival station telegraph code, arrival station bureau, arrival date, ticket type, fare, payment service serial number, mileage, user name, final status, statistical flag, transfer ticket type, creation time, etc.;

[0066] An airline ticket order information may include the following: order number, whether refunded, departure airport code, departure airport name in Chinese, departure terminal, departure city name, departure time, arrival airport code, arrival airport name in Chinese, arrival terminal, arrival city name, arrival time, duration, market carrier, market carrier flight number, date of birth, gender, transaction generation date, etc.

[0067] According to the departure and arrival times of trains and planes in the orders corresponding to each user in the autonomous intermodal transport data, the intermodal transport mode is calculated. There are two intermodal transport modes. If the arrival time of the train is earlier than the departure time of the plane, the intermodal transport mode is to take the train first and then the plane (rail-air); if the arrival time of the plane is earlier than the departure time of the train, the intermodal transport mode is to take the plane first and then the train (air-rail).

[0068] Next, the data is further processed to filter out data with the same departure point and arrival point, that is, the order data with the departure point and arrival point in the same city or station is eliminated. The same departure point and arrival point means the order is a return or round trip along the same route.

[0069] From the processed autonomous intermodal transport data and / or system intermodal transport data, tens of thousands of data are randomly selected, for example, 20,000 data are selected, information is extracted from the 20,000 data, and 20,000 sample data of intermodal transport solutions are obtained.

[0070] Each piece of sample data may include one or more of the following information: departure point (name or ID), departure point location information (such as longitude and latitude), arrival point (name or ID), arrival point location information (longitude and latitude), transfer point (name or ID), transfer point location information (such as longitude and latitude), intermodal mode label, etc. There are two types of intermodal mode labels, namely, take the railway first and then take civil aviation (railway-air) and take civil aviation first and then take the railway (air-railway).

[0071] S102: Perform normalization processing on the sample data to obtain normalized transfer point location information.

[0072] According to the departure point location information, arrival point location information and transfer point location information included in each intermodal transport plan sample data, the normalized transfer point location information is calculated.

[0073] When the departure point and the arrival point are the same, the locations of the transfer points that can be selected are different. For two plans with different departure points and arrival points, the distances between the points are different, and the directions of arrival may also be different. All directions are possible, and the transfer locations are more diverse. In the face of these spatial differences, it is difficult to directly use spatial feature information to optimize the recommended plan, and it needs to be processed into a unified dimension.

[0074] For example, Figure 2 As shown, in this embodiment, the normalization process may specifically include the following process:

[0075] S1021: Calculate the distance.

[0076] Calculating the distance includes calculating the distance between the starting point and the transfer point (first distance), the distance between the transfer point and the arrival point (second distance), and the distance between the starting point and the arrival point (third distance).

[0077] According to the location information of the departure point, the arrival point and the transfer point, for example, the location information can be longitude and latitude information, the Geopy library of Python is used to calculate the straight-line physical distance between any two of the three points. The three straight-line distances are obtained.

[0078] S1022: Determine whether a triangle is formed based on the calculated distances of the three straight lines.

[0079] Determine whether a triangle is formed based on the fact that the sum of any two sides is greater than the third side. Consider the three straight line distances as three sides and determine whether the sum of any two sides is greater than the other side. If yes, proceed to S1024; if not, proceed to step S1023.

[0080] S1023: The shortest distance among the three distances is increased by a preset length.

[0081] The three distances are regarded as three sides, and the length of the shortest side is increased by a preset length. The preset length can be a unit length, for example, when calculating in kilometers (km), 1 km is increased for the shortest side, and then the process returns to S1022 to continue to determine whether the triangle is satisfied, and if not, 1 km is increased.

[0082] In actual applications, due to certain errors in distance calculation or the use of transfer stations and transfer airports as transfer points to calculate distance, the calculated distance may not meet the triangle rule, that is, among the three distances, the sum of two distances is equal to or even less than the other distance, which is inconsistent with the actual situation. In most actual travel plans, the transfer point, departure point and arrival point are the vertices, which can form a triangle, and there is a very small probability that they will form a straight line or other shapes.

[0083] Therefore, the embodiment of the present application can solve such problems caused by errors and the like by setting S1022 and S1023, repair some abnormal sample data, and compensate for the errors by artificially extending the shortest side of the data that cannot form a triangle, so that this part of the abnormal data can also be used as sample data, enriching the data source, and the model can more fully learn the spatial characteristics of the distribution of transit points.

[0084] S1024: Normalize the departure point and the arrival point.

[0085] The starting point and the arrival point in each of the multiple sample data are fixed to two points in the rectangular coordinate system, for example, fixed to the positions of (-1, 0) and (1, 0). At this time, the side length (third side length) between the starting point and the arrival point is a fixed value of 2. The third distance between the starting point and the arrival point is divided by the side length in the coordinate system to obtain the ratio k. For example, the side length between the starting point and the arrival point in the coordinate system is 2, and the third distance is divided by 2 to obtain the ratio K.

[0086] S1025: Normalize the position of the transfer point.

[0087] The side length between the starting point and the transfer point (first distance) and the side length between the transfer point and the arrival point (second distance) are divided by K to obtain two new side lengths (first side length and second side length). In this way, the coordinates of the starting point and the arrival point in the known coordinate system are (-1,0) and (1,0) respectively, as well as the side length between the starting point and the arrival point (third side length) and the first side length between the starting point and the transfer point, and the second side length between the transfer point and the arrival point. Based on the known coordinates of the two points and the lengths of the three sides in the triangle, the coordinates of the transfer point in the coordinate system can be calculated.

[0088] It should be noted that the coordinate system may be one of the two-dimensional coordinate systems, such as a plane rectangular coordinate system, or other coordinate systems, such as a polar coordinate system.

[0089] pass Figure 2 The steps shown can normalize the departure point, arrival point and transfer point to the same coordinate system, see Figure 3 After normalizing the 20,000 sample data, the overall distribution of the normalized 20,000 transit points in the coordinate system is calculated as follows: Figure 3 shown.

[0090] observe Figure 3 The location distribution of transfer points can reveal the pattern of passengers choosing transfer points. For the transfer locations with the rail-air label (i.e., the intermodal mode is rail-air), they are mainly distributed in the upper right part of the point (-1, 0), with a small number scattered outside. For the transfer locations with the air-rail label (i.e., the intermodal mode is air-rail), they are mainly distributed in the upper left part of the point (1, 0), with a small number scattered outside. The places where transfer points are concentrated are the actual locations chosen by most passengers, and there are no obvious boundaries. It is necessary to mine the spatial feature information of this concentrated distribution area.

[0091] In this embodiment, a clustering method is adopted to mine the spatial feature information of concentrated distribution areas.

[0092] S103: Clustering the normalized location information of the plurality of transfer points to obtain an iron-empty cluster (a first cluster) and an empty-iron cluster (a second cluster).

[0093] Clustering is the process of dividing a set of data or objects into subsets with similar characteristics or attributes, so that objects in the same subset are more similar to each other, while objects in different subsets are more different. Clustering can help discover the inherent structure and patterns in data in order to better understand the data, extract useful information or conduct further analysis.

[0094] The embodiment of the present application uses a clustering method to divide the location information (or location data) of the transfer points into different clusters, and then allocates the normalized location data of the transfer points to the corresponding clusters to complete automatic classification.

[0095] Exemplarily, the clustering method used in the embodiment of the present application is a density-based spatial clustering of applications with noise (DBSCAN), which is a density-based spatial data clustering method. The algorithm uses a sufficiently dense area as the distance center, continuously grows the area, finds clusters of arbitrary shapes in noisy spatial data, connects adjacent areas with sufficiently large density, and effectively processes abnormal data.

[0096] In the embodiment of the present application, the neighborhood size parameter (eps) of DBSCAN is set to 0.08, and the cluster minimum parameter (min_samples) is set to 30.

[0097] The clustering results are as follows Figure 4 As shown in the figure, there are two clusters, namely the railway-air cluster and the air-rail cluster, and the remaining points are marked as others. The transfer points in the railway-air cluster are marked with the railway-air (first intermodal mode) label, and the transfer points in the air-rail cluster are marked with the air-rail (second intermodal mode) label. Other transfer points that are not clustered into the railway-air cluster or the air-rail cluster are marked with other labels.

[0098] S104: Using the transfer point location information with labels as training samples, training a classification model, and obtaining a trained classification model.

[0099] According to the clustering results, the classification labels in each sample were obtained, and 20,000 sample data were obtained. Using the labeled sample data as training samples, the classification model can be trained using machine learning to obtain a classification model with classification function. The trained classification model can predict the prediction scores of the three categories corresponding to the input transfer point location information.

[0100] According to the clustering results, three categories are set, namely, railway-air, railway-air, and others, which can be digitally encoded into category IDs. For example, the simplest category IDs are 0, 1, and 2. The horizontal and vertical coordinates of the transfer point in the coordinate system are selected as the input of the classification model.

[0101] The 20,000 labeled sample data were divided into a training set and a test set at a ratio of 9:1, and the training set was used to train the classification model.

[0102] Classification models can be Logistic Regression, Decision Tree, Support Vector Machine, Random Forest, etc.

[0103] In this embodiment, an XGBoost (eXtreme Gradient Boosting) model, an ensemble learning algorithm based on a gradient boosting decision tree, is selected as a classification model for training, and softmax processing is used for the prediction results.

[0104] Through experiments, the accuracy of the trained XGBoost model on the training set was 99.81%, and the accuracy on the test set was 99.65%. It can be seen that the classification effect of the model is very good, and it can excellently complete the recognition of three categories. The accurate prediction performance of the trained classification model can be used to optimize the intermodal plan and filter out unreasonable plans.

[0105] Thus, in the training phase, after the above steps S101-S104, a trained classification model is obtained. The prediction phase is described below.

[0106] Prediction stage:

[0107] S201: normalizing the intermodal transport scheme to be optimized to obtain normalized transfer point location information.

[0108] like Figure 5 As shown, the intermodal transport scheme to be optimized may be a plurality of intermodal transport schemes obtained through the recommendation module, or may be an intermodal transport scheme to be optimized obtained from other channels.

[0109] The query condition may be a query condition input or selected by a user, and the query condition includes the departure point and the arrival point information. The recommendation module may predict a variety of intermodal transport solutions to be optimized based on the departure point and the arrival point information. The optimization method proposed in the embodiment of the present application may be applied to Figure 5The optimization module shown in is used to optimize the intermodal transport plan recommended by the recommendation module.

[0110] The recommendation module can be implemented using the XGBoost model, and the optimization module can also be implemented using the XGBoost model. However, it should be noted that the two XGBoost models are different models, trained in different ways, and the trained models have different functions. The model in the recommendation module has the function of recommending intermodal solutions, and the model in the optimization module has the function of predicting categories (intermodal methods).

[0111] Specifically, the intermodal scheme to be optimized is normalized, and the normalized second transfer point position information can be obtained according to the first departure point position information, the first arrival point position information, and the first transfer point position information included in the intermodal scheme to be optimized. To prevent confusion and facilitate description, each point before normalization is defined as the first XX point, and each point after normalization is defined as the second XX point. The specific normalization method can refer to the description in the above training stage, and will not be repeated here.

[0112] S202: Inputting the second transfer point location information into a trained classification model (eg, XGBoost); obtaining predicted values ​​of the second transfer point location information relative to the three categories through the classification model.

[0113] Each recommended intermodal transport scheme is uniformly normalized and input into the classification model to obtain the prediction result [f0, f1, f2]. Among them, f0 is the prediction value of other categories, f1 is the prediction value of the railway-air category, and f2 is the prediction value of the railway-air category, and the sum of the three is 1. Alternatively, in other embodiments, f1 can be the prediction value of the railway-air category, and f2 can be the prediction value of the railway-air category.

[0114] like Figure 6 As shown in the figure, the normalized transfer point location information is input into the XGBoost model, and the model outputs a vector. After softmax processing, the vector is processed to obtain a vector with probability values ​​as its elements, such as Figure 6 The vector shown in is [0, 0.85, 0.15], which can be understood as that the predicted value of the intermodal mode of rail-air according to the transfer point location information in the current intermodal scheme to be optimized is 0.85, the probability value of the predicted intermodal mode of air-rail is 0.15, and the probability of the predicted intermodal mode of other is 0.

[0115] S203: Obtaining an initial score of the intermodal transport solution to be optimized.

[0116] The initial score of the intermodal solution to be optimized can be understood as Figure 5The initial score (i.e., recommendation score, or prediction score) predicted by the XGBoost model in the recommendation module.

[0117] The input of the model in the recommendation module includes: departure point location information, arrival point location information, passenger age, passenger gender and other fields. Specifically, the location information can be longitude and latitude information, and the selected key fields can include departure city latitude, departure city longitude, arrival city latitude, arrival city longitude, age, gender, forming a vector or array or matrix as input data in a training sample. Alternatively, in other embodiments, the field of flight departure time can also be selected, that is, the input of the model in the recommendation module can include: departure city latitude, departure city longitude, arrival city latitude, arrival city longitude, age, gender, flight departure time and other fields.

[0118] For example, suppose that after counting the transit cities in the multi-source heterogeneous data, there are 161 transit cities, that is, there are 161 transit points. In this embodiment, the two labels of transit city and intermodal mode are combined, that is, 161 transit cities are combined with two intermodal modes to obtain 322 categories, and the 322 categories are numbered respectively, for example, numbers are numbered in ascending order to correspond to the 322 categories one by one, and 322 new digital type labels are generated to replace the original two labels, and a total of 322 labels are generated, corresponding to 322 classification categories, then, the output vector of the model in the recommendation module is a 322-dimensional vector.

[0119] That is, select the latitude of the departure city, the longitude of the departure city, the latitude of the arrival city, the longitude of the arrival city, the age of the user, the gender of the user, the departure time of the flight and other information to form the input data, and input it into the XGBoost model in the trained recommendation module. The XGBoost model will output a 322-dimensional vector or array, from which the transit cities corresponding to multiple elements with probability values ​​greater than the set threshold are selected as the recommended transit cities, and the city and intermodal mode corresponding to the element are determined according to the position of the multiple elements selected in the 322-dimensional vector. Among them, the digital label corresponds to each element in the output vector one by one, and the corresponding transit city and intermodal mode can be determined according to the position of each element in the output 322-dimensional vector.

[0120] In the 322-dimensional vector predicted by the recommendation module, the value of each element is the predicted value (first predicted value) corresponding to each of the 322 categories output by the model. In this embodiment, the values ​​of each element in the 322-dimensional vector can be arranged in order from large to small, and the transit cities and corresponding intermodal modes corresponding to the top N elements are selected as the N intermodal solutions to be optimized. The predicted values ​​of the N elements in the 322-dimensional vector can be directly used as the initial score.

[0121] In other embodiments, the 322-dimensional vector may be softmax processed to obtain a normalized vector, and the normalized value may be used as the initial score. For example, N elements may be selected from the vector after softmax, and the values ​​of the normalized elements (first predicted values) of the N elements may be used as the initial scores of the corresponding N intermodal transport solutions to be optimized.

[0122] In other words, the initial score can be the predicted value of the multi-dimensional vector (e.g., 322 dimensions) predicted in the recommendation module before softmax processing, or the value after softmax processing. In this embodiment, the predicted value before softmax processing is selected as the initial score.

[0123] like Figure 7 As shown in the figure, assuming that s is the departure point and e is the arrival point, the model in the recommendation module can predict at least 6 intermodal transport solutions. The predicted transit cities are t1, t2, ... t6. The recommended 6 intermodal transport methods are as follows:

[0124] Plan 1: Transit city t1, path s→t1→e, intermodal mode: air-rail;

[0125] Plan 2: Transit city t2, path s→t2→e, intermodal mode: air-rail;

[0126] Plan 3: Transit city t3, path s→t3→e, intermodal mode: air-rail;

[0127] Plan 4: Transit city t4, path s→t4→e, intermodal mode: air-rail;

[0128] Plan 5: Transit city t5, path s→t5→e, intermodal transport mode: rail-air;

[0129] Plan 6: Transit city t6, path s→t6→e, intermodal transport mode: air-rail.

[0130] The above-recommended six intermodal transport schemes are taken as the intermodal transport schemes to be optimized. It is assumed that the values ​​of the six elements (before softmax processing) corresponding to the above-mentioned six intermodal transport schemes to be optimized in the 322-dimensional vector are 5.25, 3.78, 2.56, 2.02, 1.86, and 1.25, respectively, which are used as the initial scores of the above-mentioned six intermodal transport schemes to be optimized.

[0131] The optimization method proposed in the embodiment of the present application can be used to optimize the various intermodal transport solutions recommended by the above-mentioned recommendation module.

[0132] S204: Calculate the evaluation score of the intermodal transport solution to be optimized based on the initial score and the predicted value.

[0133] Based on the initial score and the predicted value, the evaluation score of the intermodal scheme to be optimized is calculated. Specifically, the evaluation score is calculated according to the initial score corresponding to the current intermodal scheme to be optimized and the predicted value (the predicted value obtained in S202) of the corresponding category of an intermodal mode (e.g., rail-air or air-rail) in the current intermodal scheme to be optimized output by the classification model. To prevent confusion, the predicted value (e.g., Figure 6 The value shown in ( ) is defined as the second predicted value. In other words, illustratively, the evaluation score = initial score * second predicted value. It should be noted that when calculating the evaluation score, it is necessary to select the predicted value of the category corresponding to the intermodal mode in the intermodal scheme to be optimized (the predicted value output by the classification model) for calculation. For example, if the intermodal mode in the current intermodal mode to be optimized is air-rail, then the predicted value of the air-rail category in the vector output by the classification model is used to calculate the evaluation score.

[0134] Exemplarily, a specific method of calculating the evaluation score may be to multiply the initial score output by the recommendation module and the predicted value output by the classification model, and the obtained product is the evaluation score.

[0135] For example, multiply the recommended score of the scheme before optimization by the predicted value of the transfer point of the scheme for the intermodal mode category in the scheme, that is, multiply the recommended score by f1 or f2, according to Figure 6 The above-mentioned classification method is that if the intermodal mode in the intermodal scheme before optimization is rail-air, the recommended score is multiplied by f1, and if the intermodal mode in the intermodal scheme is air-rail, the recommended score is multiplied by f2. Thus, the comprehensive evaluation score after optimization is obtained.

[0136] S205: re-rank the intermodal transport solutions to be optimized according to the evaluation scores, and filter out intermodal transport solutions with evaluation scores below a threshold.

[0137] The intermodal schemes are arranged from large to small according to the calculated evaluation scores, and schemes with scores below a score threshold are filtered out. For example, the score threshold may be set to 0.5-2.

[0138] For example, the lowest score threshold is 0.5. Assuming that the current intermodal scheme to be optimized is intermodal scheme 6 in the above example, and the intermodal mode in intermodal scheme 6 is air-rail; then, according to Figure 6 , the prediction value of the rail-air transport mode output by the classification model is 0.15. Assuming that the initial score of option 6 output by the recommendation model is 1.25 (without softmax processing), then the product of the initial score of the recommendation module and the prediction value of the classification model is 1.25. 0.15=0.1875. If the threshold of the minimum evaluation score is set to 0.5, then Solution 6 is filtered out during the optimization process.

[0139] Taking the above-mentioned plan 1 as an example, assuming that the initial score of plan 1 output by the recommendation model (without softmax processing) is 5.25, and the intermodal mode in plan 1 is air-rail, then Figure 6 As shown in the figure, the classification model, based on the transfer point location information in Scheme 1, predicts the corresponding prediction values ​​of the empty iron category in the vector to be 0.85 and 5.25 0.85=4.4625, which is higher than the score threshold of 0.5, so the solution can be retained.

[0140] The following is an example based on an actual application scenario.

[0141] Suppose a passenger wants to take air-rail transport from City S to Fuzhou. Set the filtering score to 1. When the score is greater than or equal to 1, keep the solution result, otherwise filter out the result.

[0142] After calculation, the results are shown in Table 1 below. It can be seen that the order of Plan 5 is adjusted to before Plan 4, reducing the impact of cities with large sample sizes such as Shanghai, and completing the optimization of the queue. In addition, Plan 6 is filtered out, reducing the unreasonable plan recommendations with Zhengzhou as the transit city, and completing the filtering of the plans.

[0143] Table 1

[0144]

[0145] Among them, the optimized score = initial score (before softmax) Second prediction value: In this example, the initial scores corresponding to the six solutions are: 5.173, 4.801, 4.417, 2.105, 1.679, and 1.071.

[0146] It can be seen from Table 1 above that the departure point and the transfer point can be the same city, such as Plan 2, where passengers first take a train to Daxing Airport Station, and then take a flight from Daxing International Airport in City S to Fuzhou. The score before optimization is the recommended score (initial score), and the score after optimization is the score (evaluation score) obtained by multiplying the recommended score by the predicted value of the classification model (such as f1 or f2). After optimization, the plan ranked 6th before optimization and with Zhengzhou as a transfer was filtered out, and the plan ranked 5th was optimized and ranked 4th, while the plan ranked 4th before optimization was optimized and ranked 5th. It can be seen that the optimization method proposed in the embodiment of the present application can filter out potentially unreasonable plans, and / or optimize the order of arrangement of various intermodal plans.

[0147] Thus, after optimization, 5 intermodal transport solutions from City S to Fuzhou are recommended to users. For example, in the order of recommendation, the 5 optimized intermodal transport solutions are:

[0148] Plan 1: City S → Wenzhou → Fuzhou, intermodal transport first by air and then by rail, with Wenzhou as the transit city;

[0149] Solution 2: City S → Fuzhou, intermodal transport mode is first rail and then air, and the transit city is City S; it should be noted that the method proposed in the embodiment of the present application supports the case where the departure point and the transit point are in the same city, or the transit point and the arrival point are in the same city. For example, in Solution 2, the transit point and the departure point are both in City S, which can be understood as the passenger first takes a train from one station in City S to another, and then transfers to a plane to Fuzhou, and the transit city and the departure city are both City S;

[0150] Plan 3: City S → Xiamen → Fuzhou, intermodal transport by air first and then by rail, with Xiamen as the transit city;

[0151] Plan 4: City S → Tianjin → Fuzhou, intermodal transport first by rail then by air, with Tianjin as the transit city;

[0152] Plan 5: City S → Shanghai → Fuzhou, the intermodal transport mode is air first and then rail, and the transit city is Shanghai.

[0153] According to the above exemplary description, it can be obtained that the intermodal transport scheme optimization method proposed in the embodiment of the present application can be applied to the prediction stage, such as Figure 8 As shown, the following process may be included:

[0154] S10: Obtain normalized second transfer point location information according to the first departure point location information, the first arrival point location information, and the first transfer point location information included in the intermodal transport solution to be optimized.

[0155] Among them, the intermodal transport plan to be optimized also includes intermodal transport mode information, which is one of multiple intermodal transport modes. The multiple intermodal transport modes include passengers first taking a first means of transportation and then transferring to a second means of transportation, passengers first taking the second means of transportation and then transferring to the first means of transportation, and the first means of transportation and the second means of transportation are means of transportation based on different modes of transportation.

[0156] S11: Inputting the second transfer point location information into a trained classification model; obtaining predicted values ​​of the second transfer point location information for multiple intermodal transport modes through the classification model.

[0157] S12: Obtain an initial score of the intermodal transport solution to be optimized.

[0158] S13: Based on the initial score and the predicted value, the evaluation score of the intermodal transport solution to be optimized is calculated.

[0159] S14: re-rank the intermodal transport schemes to be optimized according to the evaluation scores, and filter out intermodal transport schemes with evaluation scores below a threshold.

[0160] The present application also proposes a classification model training method, such as Fig. 9 As shown, the training method may include the following process:

[0161] S20: Acquire multiple intermodal transport solutions as sample data.

[0162] Each of the multiple intermodal transport schemes includes the first departure point location information, the first arrival point location information, the first transfer point location information, and the intermodal transport mode. For example, in the scenario of intermodal transport of rail and air, the intermodal transport mode can be rail-air or air-rail.

[0163] S21: Obtain normalized second transfer point location information according to the first departure point location information, the first arrival point location information, and the first transfer point location information.

[0164] S22: Clustering the plurality of second transfer point location information to obtain a first cluster and a second cluster.

[0165] The transfer points in the first cluster are marked with the label of the first intermodal transport mode, and the transfer points in the second cluster are marked with the label of the second intermodal transport mode; other transfer points that are not clustered into the first cluster or the second cluster are marked with the labels of other intermodal transport modes; among which, the first intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation; the second intermodal transport mode is that the passenger first takes the second means of transportation and then transfers to the first means of transportation.

[0166] S23: Using the labeled second transfer point location information as a training sample to train a classification model, thereby obtaining a trained classification model.

[0167] In some embodiments, the normalized second transit point position information is obtained based on the first starting point position information, the first arrival point position information, and the first transit point position information. Specifically, the first distance, the second distance, and the third distance are calculated based on the first starting point position information, the first arrival point position information, and the first transit point position information; then, when the sum of any two distances among the first distance, the second distance, and the third distance is greater than the other distance, it is determined that the first starting point, the first arrival point, and the first transit point form a triangle; the first starting point and the first arrival point are mapped to two points specified in the coordinate system; the distance between the first starting point and the first arrival point is K times the distance between the two specified points; the first side length and the second side length are obtained by reducing the first distance and the second distance by K times respectively; the coordinates of the transit point in the coordinate system are calculated based on the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the coordinates of the transit point obtained are used as the normalized second transit point information.

[0168] Among them, the first distance is the distance between the first starting point and the first transfer point, the second distance is the distance between the first transfer point and the first arrival point; the third distance is the distance between the first starting point and the first arrival point.

[0169] In some embodiments, after calculating the first distance, the second distance and the third distance, and before determining that the first starting point, the first arrival point and the first transfer point form a triangle, it can also be calculated that when the sum of any two distances among the first distance, the second distance and the third distance is less than or equal to another distance, that is, when it is determined that the first starting point, the first arrival point and the first transfer point do not form a triangle, the shortest distance among the first distance, the second distance and the third distance is increased by a preset length, for example, by increasing 0.5km or 1km or 1.5km or 2km at a time, etc.

[0170] It should be noted that, in other embodiments, the normalization process may not include S1022 and S1023. Fig.10 As shown, the specific process may include the following:

[0171] S1021': Calculate the distance.

[0172] S1022': Determine whether a triangle is formed, if yes, proceed to S1025', if not, proceed to S1024'.

[0173] S1023': Normalize the departure point and arrival point.

[0174] S1024': Normalize the position of the transfer point using method 2.

[0175] S1025': Normalize the position of the transfer point using method 1.

[0176] If the triangle rule is met, method one can be used for calculation, which is the method introduced in the above embodiment, that is, the coordinates of the two points in the triangle (the starting point and the arrival point) are known, and the three sides are known. The coordinates of the third point (the transfer point) can be calculated to calculate the coordinates of the transfer point in the coordinate system.

[0177] If the triangle rule is not met, the transfer point cannot be calculated according to the above triangle rule, and method 2 needs to be used for calculation. Method 2 can be that after calculating the K value, the new side length obtained by reducing the first side length or the second side length by K times is taken as the position of the transfer point on the horizontal axis, which is equal to the new side length from the starting point or the arrival point, so as to obtain the coordinates of the transfer point.

[0178] In some embodiments, clustering the plurality of second transfer point location information may be performed by using a density-based application clustering algorithm DBSCAN to cluster the plurality of second transfer point location information. The specific clustering method can be referred to the above exemplary description, which will not be described in detail here.

[0179] like Fig.11 As shown, the embodiment of the present application also proposes a multimodal transport scheme optimization device, including:

[0180] The normalization unit 1101 is used to obtain normalized second transfer point location information according to the first departure point location information, the first arrival point location information and the first transfer point location information included in the intermodal transport solution to be optimized.

[0181] The prediction unit 1102 is used to input the second transfer point location information into a trained classification model; and obtain a prediction value of the second transfer point location information relative to the category corresponding to the first intermodal transport mode through the classification model.

[0182] The acquisition unit 1103 is used to acquire the initial score of the intermodal transport solution to be optimized.

[0183] The evaluation unit 1104 is used to calculate the evaluation score of the intermodal transport scheme to be optimized based on the initial score and the predicted value; and re-rank the intermodal transport scheme to be optimized according to the evaluation score, and filter out the intermodal transport scheme with an evaluation score lower than a threshold.

[0184] like Fig.12 As shown, the embodiment of the present application also proposes a classification model training device, including:

[0185] The acquisition unit 1201 is used to acquire multiple intermodal transport schemes as sample data; each of the multiple intermodal transport schemes includes first departure point location information, first arrival point location information, first transfer point location information and a first intermodal transport mode.

[0186] The normalization unit 1202 is used to obtain normalized second transfer point position information according to the first departure point position information, the first arrival point position information, and the first transfer point position information.

[0187] The clustering unit 1203 is used to cluster the plurality of second transfer point location information to obtain a first cluster and a second cluster.

[0188] The training unit 1204 is used to use the second transfer point location information with a label as a training sample to train the classification model to obtain a trained classification model.

[0189] In some embodiments, the normalization unit can be specifically used to execute the above-mentioned normalization process, for example, it can be specifically used to execute the following process: calculate the first distance, the second distance and the third distance according to the first starting point position information, the first arrival point position information and the first transit point position information; determine that the first starting point, the first arrival point and the first transit point form a triangle when the sum of any two distances among the first distance, the second distance and the third distance is greater than the other distance; map the first starting point and the first arrival point to two points specified in the coordinate system; the distance between the first starting point and the first arrival point is K times the distance between the two specified points; reduce the first distance and the second distance by K times respectively to obtain the first side length and the second side length; calculate the coordinates of the transit point in the coordinate system according to the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the coordinates of the transit point obtained are used as the normalized second transit point information.

[0190] In some embodiments, after calculating the first distance, the second distance, and the third distance, and before determining that the first starting point, the first arrival point, and the first transit point form a triangle, the normalization unit is further used to calculate that when the sum of any two distances among the first distance, the second distance, and the third distance is less than or equal to the other distance, it is determined that the first starting point, the first arrival point, and the first transit point do not form a triangle; and the shortest distance among the first distance, the second distance, and the third distance is increased by a preset unit length.

[0191] In some embodiments, when clustering the plurality of second transfer point location information, the clustering unit is specifically configured to cluster the plurality of second transfer point location information using a density-based application clustering algorithm DBSCAN.

[0192] In summary, the embodiment of the present application proposes an algorithm for optimizing and filtering air-rail transport recommendation plans based on spatial characteristics. The method uses the location data of passengers' actual choices as the basis, and then designs a unified normalization method to facilitate data analysis and application. Furthermore, the clustering and classification methods of machine learning are used to complete the classification of transfer locations. Finally, the transport plan is re-evaluated through the predicted value output by the classification model, thereby achieving the optimization and filtering of air-rail transport recommendation plans.

[0193] The embodiment of the present application uses spatial feature information to further adjust each recommended plan, automatically adapt and adjust the transit city plan, thereby completing the optimization and filtering of the recommended air-rail transport plans, making the recommended air-rail transport plans more reliable.

[0194] An embodiment of the present application further provides an electronic device, the electronic device comprising: a processor, the processor being used to execute a computer program or instruction in a memory to implement a method as described in any of the above embodiments.

[0195] The electronic device involved in the present application may be one or more of the following devices: a smart phone, a portable computer (Tablet Personal Computer, Tablet PC), a laptop computer (laptop), a desktop computer (Desktop computer), a wearable device, an extended reality (extended reality, XR) device such as augmented reality (AR), virtual reality (VR), mixed reality (MR), an ultra-mobile personal computer (UMPC), a netbook or a personal digital assistant (PDA), etc. The embodiments of the present application do not impose any special restrictions on the specific type of the electronic device.

[0196] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in this application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk SolidState Disk), etc.

[0197] In the embodiments of the present application, "at least one" refers to one or more, and "plurality" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0198] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A classification model training method, characterized in that: The method is applied to an electronic device, and the method comprises: Acquire multiple intermodal transport schemes as sample data; each of the multiple intermodal transport schemes includes first departure point location information, first arrival point location information, first transfer point location information, and an intermodal transport mode; Calculate a first distance, a second distance and a third distance according to the first departure point location information, the first arrival point location information and the first transfer point location information; wherein the first distance is the distance between the first departure point and the first transfer point, the second distance is the distance between the first transfer point and the first arrival point; and the third distance is the distance between the first departure point and the first arrival point; When the sum of any two of the first distance, the second distance and the third distance is greater than the other distance, determining that the first departure point, the first arrival point and the first transfer point form a triangle; Mapping the first departure point and the first arrival point to two specified points in a coordinate system; the distance between the first departure point and the first arrival point is K times the distance between the two specified points; The first distance and the second distance are respectively reduced by K times to obtain a first side length and a second side length; Calculate the coordinates of the transfer point in the coordinate system according to the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the obtained coordinates of the transfer point are used as the normalized second transfer point position information; Clustering the location information of multiple second transfer points to obtain a first cluster and a second cluster; the transfer points in the first cluster are marked with a label of the first intermodal transport mode, and the transfer points in the second cluster are marked with a label of the second intermodal transport mode; other transfer points that are not clustered into the first cluster or the second cluster are marked with other labels; wherein the first intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation; and the second intermodal transport mode is that the passenger first takes the second means of transportation and then transfers to the first means of transportation; The second transfer point location information with the label is used as a training sample to train the classification model to obtain a trained classification model.

2. The method according to claim 1, characterized in that After calculating the first distance, the second distance, and the third distance, and before determining that the first departure point, the first arrival point, and the first transfer point form a triangle, the method further includes: When the sum of any two of the first distance, the second distance, and the third distance is less than or equal to another distance, it is determined that the first departure point, the first arrival point, and the first transfer point do not form a triangle; The shortest distance among the first distance, the second distance and the third distance is increased by a preset length.

3. The method according to claim 1 or 2, characterized in that: Clustering the location information of the plurality of second transfer points includes: The density-based application clustering algorithm DBSCAN is used to cluster the location information of multiple second transfer points.

4. A method for optimizing an intermodal transport scheme, characterized in that: The method is applied to an electronic device, and the method comprises: According to the first departure point location information, the first arrival point location information and the first transfer point location information included in the intermodal transport scheme to be optimized, the normalized second transfer point location information is obtained; the intermodal transport scheme to be optimized also includes an intermodal transport mode; wherein the intermodal transport mode is one of multiple intermodal transport modes, wherein the multiple intermodal transport modes include a passenger first taking a first means of transport and then transferring to a second means of transport, or a passenger first taking a second means of transport and then transferring to the first means of transport; the first means of transport and the second means of transport are means of transport based on different modes of transport; Inputting the second transfer point location information into a trained classification model; obtaining predicted values ​​of the second transfer point location information for multiple intermodal transport modes through the classification model; wherein the trained classification model is trained using the method described in any one of claims 1 to 3; Obtaining an initial score of the intermodal transport solution to be optimized; Calculating an evaluation score of the intermodal transport solution to be optimized based on the initial score and the predicted value; The intermodal transport schemes to be optimized are reordered according to the evaluation scores, and intermodal transport schemes with evaluation scores below a threshold are filtered out.

5. A classification model training device, characterized in that: The device comprises: An acquisition unit, configured to acquire a plurality of intermodal transport schemes as sample data; each of the plurality of intermodal transport schemes includes first departure point location information, first arrival point location information, first transfer point location information, and a first intermodal transport mode; A normalization unit is used to calculate a first distance, a second distance and a third distance according to the first starting point location information, the first arrival point location information and the first transfer point location information; wherein the first distance is the distance between the first starting point and the first transfer point, the second distance is the distance between the first transfer point and the first arrival point; the third distance is the distance between the first starting point and the first arrival point; when the sum of any two of the first distance, the second distance and the third distance is greater than the other distance, it is determined that the first starting point, the first arrival point and the first transfer point form a triangle; the first starting point and the first arrival point are mapped to two points specified in the coordinate system; the distance between the first starting point and the first arrival point is K times the distance between the two specified points; the first side length and the second side length are obtained by reducing the first distance and the second distance by K times respectively; the coordinates of the transfer point in the coordinate system are calculated according to the coordinates of the two specified points in the coordinate system and the first side length and the second side length; the coordinates of the transfer point obtained are used as the normalized second transfer point location information; A clustering unit is used to cluster the location information of multiple second transfer points to obtain a first cluster and a second cluster; the transfer points in the first cluster are marked with a label of a first intermodal transport mode, and the transfer points in the second cluster are marked with a label of a second intermodal transport mode; other transfer points that are not clustered into the first cluster or the second cluster are marked with labels of other intermodal transport modes; wherein the first intermodal transport mode is that the passenger first takes the first means of transportation and then transfers to the second means of transportation; and the second intermodal transport mode is that the passenger first takes the second means of transportation and then transfers to the first means of transportation; The training unit is used to use the second transfer point location information with a label as a training sample to train the classification model and obtain a trained classification model.

6. The device according to claim 5, characterized in that After calculating the first distance, the second distance, and the third distance, and before determining that the first departure point, the first arrival point, and the first transfer point form a triangle, the normalization unit is further used to: When the sum of any two of the first distance, the second distance, and the third distance is less than or equal to another distance, it is determined that the first departure point, the first arrival point, and the first transfer point do not form a triangle; The shortest distance among the first distance, the second distance and the third distance is increased by a preset unit length.

7. The device according to claim 5 or 6, characterized in that When clustering the plurality of second transfer point location information, the clustering unit is specifically used to: The density-based application clustering algorithm DBSCAN is used to cluster the location information of multiple second transfer points.

8. A device for optimizing intermodal transport scheme, characterized in that: The device comprises: a normalization unit, configured to obtain normalized second transfer point location information according to first departure point location information, first arrival point location information, and first transfer point location information included in the intermodal scheme to be optimized; the intermodal scheme to be optimized also includes an intermodal mode; wherein the intermodal mode is that a passenger first takes a first means of transportation and then transfers to a second means of transportation, or that a passenger first takes a second means of transportation and then transfers to the first means of transportation; the first means of transportation and the second means of transportation are means of transportation based on different modes of transportation; A prediction unit, configured to input the second transfer point location information into a trained classification model; and obtain a prediction value of the second transfer point location information relative to a category corresponding to the first intermodal transport mode through the classification model; wherein the trained classification model is trained using the device according to any one of claims 5 to 7; An acquisition unit, configured to acquire an initial score of the intermodal transport solution to be optimized; An evaluation unit calculates an evaluation score of the intermodal transport scheme to be optimized based on the initial score and the predicted value; and, according to the evaluation score, reorders the intermodal transport scheme to be optimized and filters out intermodal transport schemes with evaluation scores below a threshold.

9. An electronic device, characterized in that: The electronic device comprises: A processor, wherein the processor is configured to execute a computer program or instruction in a memory to implement the method according to any one of claims 1 to 3 or 4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program implements the method according to any one of claims 1-3 or 4 when executed by a processor.

11. A computer program product, characterized in that The computer program product includes a program, and when the program is executed by an electronic device, the electronic device implements the method according to any one of claims 1-3 or 4.

Images