Travel path planning method and device based on recommendation principle, and storage medium

Abstract

The invention relates to a travel path planning method and device based on a recommendation principle and a storage medium. According to the method, a user provides user characteristics of a current trip; extracting all routes between the starting point and the target point, and splitting the all routes into a plurality of sub-routes according to the overall planning of the road; extracting features of roads where the sub-routes are located and vehicle travel features on the sub-routes from the traffic attribute data; initial embedded representation representing user objects and features thereof, sub-routes and road features thereof, and vehicle travel features on the sub-routes are constructed through a pre-trained embedded layer and pooling layer; carrying out graph convolution extraction on the initial embedding representation of the vehicle travel characteristics and the road characteristics to obtain association between the two so as to construct an initial embedding representation of a travel mode; and obtaining a similarity score between the travel mode initial embedded representation and the user object of the current travel and the initial embedded representation of the characteristics of the user object according to the travel mode initial embedded representation and the user object, and making a decision according to the similarity score between the to-be-selected travel mode and the user during path planning.

Description

Technical Field

[0001] The present invention relates to the technical field of route planning, and in particular to a travel route planning method based on a recommendation principle. Background Art

[0002] At present, traditional navigation is often planned based on the optimal distance between the starting point and the target point. For self-driving, many route plans determined by navigation cannot meet the navigation needs of users. In addition to distance, there are many other factors that affect navigation, such as vehicle factors, user factors, and other road factors besides distance. For example, with the development of electric vehicles, their charging needs are an important factor affecting navigation during long-distance travel. Among two optional routes of the same length, in order to meet the charging needs, a route with more service areas will be selected, and different drivers tend to take routes with more service areas; when users travel for leisure travel and self-driving, the distribution of scenic spots around the road will affect the choice of route, so a route planning method that better meets user needs is needed. Summary of the invention

[0003] In order to solve the above technical problem or at least partially solve the above technical problem, the present invention provides a travel route planning method based on the recommendation principle.

[0004] In a first aspect, the present invention provides a travel route planning method based on a recommendation principle, comprising:

[0005] Road side collects travel data to obtain original travel big data, and divides travel big data into user attribute data and traffic attribute data according to traffic travel scenarios;

[0006] The user provides the user characteristics of the current trip; all routes between the starting point and the destination are extracted, and all routes are divided into several sub-routes according to the overall road planning; the road characteristics of the sub-routes are extracted from the traffic attribute data, as well as the vehicle travel characteristics on the sub-routes under different time periods and climate conditions;

[0007] The pre-trained embedding layer and pooling layer are used to construct an initial embedding representation of the same size that represents the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route;

[0008] Perform graph convolution on the initial embedding representation of vehicle travel characteristics and road characteristics to extract the relationship between them to construct the initial embedding representation of travel mode;

[0009] The similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its characteristics of the current travel is obtained. During path planning, the path and vehicle travel characteristics are selected based on the similarity score between the selected travel mode and the user.

[0010] Furthermore, the user attribute data includes the following contents:

[0011] The data of the vehicle driven by the user, including: vehicle brand, vehicle type, energy type and city of ownership specified by the license plate, registered address, purchase time, and vehicle fuel consumption;

[0012] User identity factors include: user gender, age, native place, driving experience, education level, income, travel motivation such as travel, work, visiting relatives, travel preferences such as self-driving, leasing, borrowing, sharing, first address, second address;

[0013] User occupation attribute: industry.

[0014] Furthermore, the traffic attribute data includes the following contents:

[0015] Basic road data, including: road construction time, number of road lanes, lane width, national inspection score, design flow, overhaul time, and maintenance frequency;

[0016] Distribution data of roads and surrounding infrastructure, including: distribution of towns, scenic spots, and service areas within 100 kilometers of roads, number of transportation hubs, tunnels, toll booths, and ETC lanes along the roads, roadside signs, information boards, cameras, radars, and antenna facilities;

[0017] Road traffic data, including: historical average traffic volume, truck traffic volume, time per 100 kilometers in different time periods, and road section toll levels;

[0018] Road traffic policy data, including: traffic restrictions, driving bans, speed limits, and reconstruction and expansion;

[0019] Vehicle travel data, including: historical toll data, travel frequency, driving mileage, toll amount, payment method, travel vehicle, travel time, entrance and exit, navigation reference route and detour route, average speed; among them, the increased cost of the detour route relative to the navigation reference route is provided, and the increased cost is represented by the function c(Δs, Δt), where Δs is the relative increase in distance and Δt is the estimated relative increase in time;

[0020] Vehicle travel time elements, including: historical departure date, departure time, average travel time, travel direction, holiday characteristics, weekend and weekday characteristics;

[0021] Weather factors for vehicle travel include: historical weather of the road section, local climate of the road, characteristics of rivers, lakes, wetlands and bridges, and sunlight time and direction.

[0022] Furthermore, before extracting features, the travel big data is first clustered to reduce the data dimensions of user objects and their features, road features of sub-routes, and vehicle travel features on sub-routes under different time periods and climate conditions.

[0023] Furthermore, the process of extracting the association between vehicle travel characteristics and road characteristics to construct the initial embedding representation of travel mode includes:

[0024] Construct the interaction matrix between vehicle travel characteristics and road characteristics based on the original travel data history The interaction matrix between vehicle travel characteristics and road characteristics defines the first adjacency matrix between vehicle travel characteristics and road characteristics:

[0025]

[0026] Then the vehicle travel characteristics and road characteristics are embedded and propagated iteratively through the first adjacency matrix A1:

[0027]

[0028] Among them, D1 is the degree matrix of the adjacency matrix A1, which is in the form of a diagonal matrix. is the embedding representation matrix of the sub-route and its road features R, the vehicle travel features V on the sub-route, and j is the number of iterations.

[0029] Aggregate the embedding representation propagation results of all sub-routes and vehicle travel characteristics to obtain the aggregation result:

[0030]

[0031] Among them, β0,β1,...,β J is the weight in the weighted aggregation process;

[0032] The aggregation result E R,V Projected into the initial embedding representation of travel mode through linear mapping

[0033] Furthermore, the obtaining of the similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its features of the current travel includes: constructing an interaction matrix M2∈R between the historical user objects and the travel modes according to the big travel data X×Y , the interaction matrix between user objects and travel modes is used to define the second adjacency matrix that represents the graph relationship between users and travel modes:

[0034]

[0035] Then the user objects and travel modes are embedded and propagated iteratively through the second adjacency matrix A2:

[0036]

[0037] Among them, D2 is the degree matrix of the adjacency matrix A2, which is in the form of a diagonal matrix. is the embedding representation matrix composed of the embedding representation of the user object and its features and the embedding of the travel mode, and k is the number of iterations;

[0038] Aggregate all the embedding propagation results. In the specific implementation process, this application adopts weighted aggregation:

[0039] E U,C =γ0E 0 +γ1E 1 +...+γ K E K ;

[0040] Among them, γ0,γ1,...,γ J To add weights to the weighted aggregation process, the final user embedding matrix E is decomposed from the aggregation result. U and the travel mode embedding matrix E C ;

[0041] The similarity score is calculated using the inner product of the final user and travel mode embedding representations.

[0042] Furthermore, when training the model, the loss function used for training is as follows:

[0043]

[0044] For a given positive sample, sample a set of negative samples R n ,V n To minimize the probability of negative samples, C(r p ,v p ) is the positive route r preferred by user u p and the positive vehicle travel feature v preferred by user u p is the combination of user u’s preferred travel mode, s(,) is the similarity function for the corresponding object embedding representation, C(r p ,v p ) is the negative sub-route r abandoned by user u n and the negative vehicle travel characteristics v abandoned by user u n is the combination of , that is, the travel mode item abandoned by user u, and α is the set hyper parameter.

[0045] Furthermore, the meaning of the loss function is:

[0046]

[0047] make

[0048]

[0049] but

[0050] The loss function is The upper bound of .

[0051] In a second aspect, the present invention provides a travel route planning device based on the recommendation principle, comprising: at least one processing unit, the processing unit is connected to a storage unit via a bus unit, the storage unit stores a computer program, and when the computer program is executed by the processing unit, the travel route planning method based on the recommendation principle is implemented.

[0052] In a third aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the travel route planning method based on the recommendation principle is implemented.

[0053] The above technical solution provided by the embodiment of the present invention has the following advantages compared with the prior art:

[0054] In order to integrate more factors that determine the path planning decision into the path planning process, the present invention extracts user objects and their characteristics from user attribute data; extracts all routes between the starting point and the target point, and divides all routes into several sub-routes according to the overall road planning; extracts the road characteristics of the sub-routes from the traffic attribute data, as well as the vehicle travel characteristics on the sub-routes under different time periods and climate conditions. The association between the initial embedding representation of vehicle travel characteristics and road characteristics is extracted by graph convolution to construct the initial embedding representation of the travel mode; the similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its characteristics of the current trip is obtained, and when planning the path, the path and vehicle travel characteristics are selected according to the similarity score between the selected travel mode and the user. Based on the ranking of the similarity scores between the user and the embedded representation of the travel mode, the path planning that better meets the user's current needs after considering many factors is recommended. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0057] Figure 1A flow chart of a travel route planning method based on a recommendation principle provided in an embodiment of the present invention;

[0058] Figure 2 An architecture diagram of a travel route planning method network based on a recommendation principle provided by an embodiment of the present invention;

[0059] Figure 3 A schematic diagram of a travel route planning device based on a recommendation principle provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0062] Example 1

[0063] The technology of the present invention implements a travel route planning method based on the recommendation principle, including:

[0064] Training phase:

[0065] Road parties collect travel data to obtain original travel big data.

[0066] In this application, according to the traffic travel scenario, the travel big data is classified into two categories: one is user attribute data, which is mostly static data, and the other is traffic attribute data, which is mostly dynamically changing data. In the traffic travel scenario, user attribute data is often much more than traffic broadcast data.

[0067] User attribute data in the transportation scenario includes the following:

[0068] The data of the vehicle driven by the user, including: vehicle brand, vehicle type, energy type and city of ownership specified by the license plate, registered address, purchase time, and vehicle fuel consumption;

[0069] User identity factors include: user gender, age, native place, driving experience, education level, income, travel motivation such as travel, work, visiting relatives, travel preferences such as self-driving, leasing, borrowing, sharing, first address, second address;

[0070] User occupation attribute: industry.

[0071] Traffic attribute data in traffic travel scenarios include the following:

[0072] Basic road data, including: road construction time, number of road lanes, lane width, national inspection score, design flow, overhaul time, and maintenance frequency;

[0073] Distribution data of roads and surrounding infrastructure, including: distribution of towns, scenic spots, and service areas within 100 kilometers of roads, number of transportation hubs, tunnels, toll booths, and ETC lanes along the roads, roadside signs, information boards, cameras, radars, and antenna facilities;

[0074] Road traffic data, including: historical average traffic volume, truck traffic volume, time per 100 kilometers in different time periods, and road section toll levels;

[0075] Road traffic policy data, including: traffic restrictions, driving bans, speed limits, and reconstruction and expansion;

[0076] Vehicle travel data, including: historical toll data, travel frequency, driving mileage, toll amount, payment method, travel vehicle, travel time, entrance and exit, navigation reference route and detour route, average speed; among them, the increased cost of the detour route relative to the navigation reference route is provided, and the increased cost is represented by the function c(Δs, Δt), where Δs is the relative increase in distance and Δt is the estimated relative increase in time;

[0077] Vehicle travel time elements, including: historical departure date, departure time, average travel time, travel direction, holiday characteristics, weekend and weekday characteristics;

[0078] Weather factors for vehicle travel include: historical weather of the road section, local climate of the road, characteristics of rivers, lakes, wetlands and bridges, and sunlight time and direction.

[0079] With the accumulation of traffic travel scenarios, a large amount of historical user, vehicle, and road-related data will be generated. The path navigation between the starting point and the destination point is often limited to the optimal path, and other factors will not be considered except the road conditions. The travel big data of this application can provide planning decision factors adapted to user, vehicle data, and road conditions for path planning.

[0080] In order to integrate these decision factors into the route planning process, this application extracts user objects and their features from user attribute data; extracts maps and divides all routes into several sub-routes according to the overall road plan; extracts the road features of the sub-routes from traffic attribute data, as well as the vehicle travel features on the sub-routes at different time periods and climate conditions. Before extracting features, the travel big data is first clustered to reduce the data dimensions of user objects and their features, road features of the sub-routes, and vehicle travel features on the sub-routes at different time periods and climate conditions.

[0081] like Figure 2 As shown in the figure, the embedding layer and the pooling layer are used to construct the initial embedding representation of the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route, which are of the same size. and represents, where:

[0082]

[0083]

[0084] pool1(), pool2(), pool3() represent the pooling functions implemented by the pooling layer, and embedding() is the embedding layer. The embedding layer maps the user object and its features, the sub-route and its road features, and the vehicle travel features on the sub-route to a low-dimensional initial embedding representation, and vectorizes the three.

[0085] The association between vehicle travel characteristics and road characteristics is extracted to construct the travel mode embedding representation. In order to capture the relationship between vehicle travel characteristics and road characteristics, the interaction matrix between vehicle travel characteristics and road characteristics is constructed based on the original travel data. The interaction matrix between vehicle travel characteristics and road characteristics defines the first adjacency matrix between vehicle travel characteristics and road characteristics:

[0086]

[0087] Then the vehicle travel characteristics and road characteristics are embedded and propagated iteratively through the first adjacency matrix A1:

[0088]

[0089] Among them, D1 is the degree matrix of the adjacency matrix A1, which is in the form of a diagonal matrix. is the embedding representation matrix of the sub-route and its road features R, the vehicle travel features V on the sub-route, and j is the number of iterations.

[0090] Aggregate the embedding representation propagation results of all sub-routes and vehicle travel characteristics to obtain the aggregation result:

[0091]

[0092] Among them, β0,β1,...,β J In order to add weights to the weighted aggregation process, each layer of embedding representation matrix is ​​evolved based on the initial embedding matrix and the first adjacency matrix. As the level goes deeper, the order of the first adjacency matrix increases. Therefore, the embedding propagation aggregation process actually aggregates the different orders of adjacency relationships of the initial embedding representation matrix together. For each vehicle travel feature, the different orders of embeddings of all sub-routes and road features adjacent to each vehicle travel feature are integrated into the embedding of the vehicle travel feature; for each sub-route and its road feature, the different orders of embeddings of all sub-routes and road features adjacent to each vehicle travel feature are integrated into the embedding of the vehicle travel feature, so that the connection between the vehicle travel feature and the sub-route and its road feature can be effectively established.

[0093] The aggregation result E R,V Projected into the initial embedding representation of travel mode through linear mapping

[0094] Construct the interaction matrix M2∈R between historical user objects and travel modes based on big travel data X×Y , the interaction matrix between user objects and travel modes is used to define the second adjacency matrix that represents the graph relationship between users and travel modes:

[0095]

[0096] Then the user objects and travel modes are embedded and propagated iteratively through the second adjacency matrix A2:

[0097]

[0098] Among them, D2 is the degree matrix of the adjacency matrix A2, which is in the form of a diagonal matrix. is the embedding representation matrix composed of the embedding representation of the user object and its features and the travel mode embedding, and k is the number of iterations.

[0099] Aggregate all the embedding propagation results. In the specific implementation process, this application adopts weighted aggregation:

[0100] E U,C =γ0E 0 +γ1E 1 +...+γ K E K ;

[0101] Among them, γ0,γ1,...,γ J In order to add weights to the weighted aggregation process, each layer of embedding matrices is evolved based on the initial embedding matrix and the adjacency matrix. As the level goes deeper, the order of the adjacency matrix increases. The embedding propagation aggregation process is actually aggregating together the different orders of adjacency relationships of the initial embedding matrix. For each user, the different orders of embeddings of all the travel mode items adjacent to each user are integrated into the user embedding; for each travel mode item, the different orders of embeddings of all the users adjacent to each travel mode item are integrated into the embedding of the travel mode, so that the preference tendency between users and travel modes can be effectively established.

[0102] Decompose the final user embedding matrix E from the aggregation results U and the travel mode embedding matrix E C .

[0103] The similarity score is obtained by using the inner product of the final user and travel mode embedding representation, and the similarity score matrix is:

[0104]

[0105] When training the model, the loss function used for training is as follows:

[0106]

[0107] In order to deal with the problem that it is impossible to perform probability statistics on all negative samples in large-scale data, in this application, for a given positive sample, a set of negative samples R n ,V n To minimize the probability of negative samples. Among them, C(r p ,v p ) is the positive route r preferred by user u p and the positive vehicle travel feature v preferred by user u p is the combination of user u’s preferred travel mode, s(,) is the similarity function for the corresponding object embedding representation, C(r p ,v p ) is the negative sub-route r abandoned by user u n and the negative vehicle travel characteristics v abandoned by user u n The combination of is the travel mode item abandoned by user u, and α is the set hyperparameter. The meaning of loss function is:

[0108]

[0109] make

[0110]

[0111] but

[0112] The loss function is The upper bound of ; It can effectively supervise the model so that the model maximizes the similarity between the user and the preferred travel mode, minimizes the similarity between the user and the abandoned travel mode, and is simpler in operation.

[0113] like Figure 1 As shown, the application phase process includes:

[0114] The user provides the user characteristics of the current trip;

[0115] All routes between the starting point and the target point are extracted, and all routes are divided into several sub-routes according to the overall road planning; all routes between the starting point and the target point include: navigation routes, detour routes with increased costs within a set cost threshold compared to the navigation routes.

[0116] Extract the road characteristics of the sub-route from the traffic attribute data, as well as the vehicle travel characteristics on the sub-route under different time periods and climate conditions;

[0117] The pre-trained embedding layer and pooling layer are used to construct an initial embedding representation of the same size that represents the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route;

[0118] The relationship between vehicle travel characteristics and the initial embedding representation of road characteristics is extracted by graph convolution to construct the initial embedding representation of travel mode.

[0119] Obtain a similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its features of the current travel;

[0120] When planning a route, the route and vehicle travel characteristics are selected based on the similarity scores between the selected travel modes and users.

[0121] Example 2

[0122] See also Figure 3 As shown, an embodiment of the present invention provides a travel path planning device based on the recommendation principle, including: at least one processing unit, the processing unit is connected to a storage unit through a bus unit, the storage unit is a computer-readable storage medium, and can be used to store software programs, computer executable programs and modules, such as the software programs, computer executable programs and modules corresponding to a travel path planning method based on the recommendation principle in an embodiment of the present invention. The processing unit implements the above-mentioned travel path planning method based on the recommendation principle by running the software programs, computer executable programs and modules stored in the storage unit, including:

[0123] Road side collects travel data to obtain original travel big data, and divides travel big data into user attribute data and traffic attribute data according to traffic travel scenarios;

[0124] The user provides the user characteristics of the current trip; all routes between the starting point and the destination are extracted, and all routes are divided into several sub-routes according to the overall road planning; the road characteristics of the sub-routes are extracted from the traffic attribute data, as well as the vehicle travel characteristics on the sub-routes under different time periods and climate conditions;

[0125] The pre-trained embedding layer and pooling layer are used to construct an initial embedding representation of the same size that represents the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route;

[0126] Perform graph convolution on the initial embedding representation of vehicle travel characteristics and road characteristics to extract the relationship between them to construct the initial embedding representation of travel mode;

[0127] The similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its characteristics of the current travel is obtained. During path planning, the path and vehicle travel characteristics are selected based on the similarity score between the selected travel mode and the user.

[0128] Of course, the computer program stored in the storage unit of a travel route planning device based on the recommendation principle provided in an embodiment of the present invention is not limited to the method operations described above, and can also execute related operations in a travel route planning method based on the recommendation principle provided in any embodiment of the present invention.

[0129] Example 3

[0130] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed, the travel path planning method based on the recommendation principle is implemented, including:

[0131] Road side collects travel data to obtain original travel big data, and divides travel big data into user attribute data and traffic attribute data according to traffic travel scenarios;

[0132] The user provides the user characteristics of the current trip; all routes between the starting point and the destination are extracted, and all routes are divided into several sub-routes according to the overall road planning; the road characteristics of the sub-routes are extracted from the traffic attribute data, as well as the vehicle travel characteristics on the sub-routes under different time periods and climate conditions;

[0133] The pre-trained embedding layer and pooling layer are used to construct an initial embedding representation of the same size that represents the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route;

[0134] Perform graph convolution on the initial embedding representation of vehicle travel characteristics and road characteristics to extract the relationship between them to construct the initial embedding representation of travel mode;

[0135] The similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its characteristics of the current travel is obtained. During path planning, the path and vehicle travel characteristics are selected based on the similarity score between the selected travel mode and the user.

[0136] A computer-readable storage medium provided in an embodiment of the present invention stores a computer program which is not limited to the method operations described above, but can also execute related operations in a travel route planning method based on a recommendation principle provided in any embodiment of the present invention.

[0137] In the embodiments provided by the present invention, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the structural embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, structures or units, which can be electrical, mechanical or other forms.

[0138] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0139] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0140] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A travel route planning method based on recommendation principle, characterized in that: include: Road side collects travel data to obtain original travel big data, and divides travel big data into user attribute data and traffic attribute data according to traffic travel scenarios; The user provides the user characteristics of the current trip; all routes between the starting point and the destination are extracted, and all routes are divided into several sub-routes according to the overall road planning; the road characteristics of the sub-routes are extracted from the traffic attribute data, as well as the vehicle travel characteristics on the sub-routes under different time periods and climate conditions; The pre-trained embedding layer and pooling layer are used to construct an initial embedding representation of the same size that represents the user object and its features U, the sub-route and its road features R, and the vehicle travel features V on the sub-route; Perform graph convolution on the initial embedding representation of vehicle travel characteristics and road characteristics to extract the relationship between them to construct the initial embedding representation of travel mode; The similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its characteristics of the current travel is obtained. During path planning, the path and vehicle travel characteristics are selected based on the similarity score between the selected travel mode and the user.

2. The travel route planning method based on the recommendation principle according to claim 1 is characterized in that: The user attribute data includes the following contents: The data of the vehicle driven by the user, including: vehicle brand, vehicle type, energy type and city of ownership specified by the license plate, registered address, purchase time, and vehicle fuel consumption; User identity factors include: user gender, age, native place, driving experience, education level, income, travel motivation such as travel, work, visiting relatives, travel preferences such as self-driving, leasing, borrowing, sharing, first address, second address; User occupation attribute: industry.

3. The travel route planning method based on the recommendation principle according to claim 1, characterized in that: The traffic attribute data includes the following contents: Basic road data, including: road construction time, number of road lanes, lane width, national inspection score, design flow, overhaul time, and maintenance frequency; Distribution data of roads and surrounding infrastructure, including: distribution of towns, scenic spots, and service areas within 100 kilometers of roads, number of transportation hubs, tunnels, toll booths, and ETC lanes along the roads, roadside signs, information boards, cameras, radars, and antenna facilities; Road traffic data, including: historical average traffic volume, truck traffic volume, time per 100 kilometers in different time periods, and road section toll levels; Road traffic policy data, including: traffic restrictions, driving bans, speed limits, and reconstruction and expansion; Vehicle travel data, including: historical toll data, travel frequency, driving mileage, toll amount, payment method, travel vehicle, travel time, entrance and exit, navigation reference route and detour route, average speed; among them, the increased cost of the detour route relative to the navigation reference route is provided, and the increased cost is represented by the function c(Δs, Δt), where Δs is the relative increase in distance and Δt is the estimated relative increase in time; Vehicle travel time elements, including: historical departure date, departure time, average travel time, travel direction, holiday characteristics, weekend and weekday characteristics; Weather factors for vehicle travel include: historical weather of the road section, local climate of the road, characteristics of rivers, lakes, wetlands and bridges, and sunlight time and direction.

4. The travel route planning method based on the recommendation principle according to claim 1, characterized in that: Before extracting features, the travel big data is first clustered to reduce the data dimensions of user objects and their features, road features of sub-routes, and vehicle travel features on sub-routes under different time periods and climate conditions.

5. The travel route planning method based on the recommendation principle according to claim 1, characterized in that: The construction of the initial embedding representation of the travel mode by performing graph convolution extraction on the initial embedding representation of the vehicle travel characteristics and the road characteristics includes: Construct the interaction matrix between vehicle travel characteristics and road characteristics based on the original travel data history The interaction matrix between vehicle travel characteristics and road characteristics defines the first adjacency matrix between vehicle travel characteristics and road characteristics: Then the vehicle travel characteristics and road characteristics are embedded and propagated iteratively through the first adjacency matrix A1: Among them, D1 is the degree matrix of the adjacency matrix A1, which is in the form of a diagonal matrix. is the embedding representation matrix of the sub-route and its road features R, the vehicle travel features V on the sub-route, and j is the number of iterations. Aggregate the embedding representation propagation results of all sub-routes and vehicle travel characteristics to obtain the aggregation result: Among them, β0,β1,...,β J is the weight in the weighted aggregation process; The aggregation result E R,V Projected into the initial embedding representation of travel mode through linear mapping 6. The travel route planning method based on recommendation principle according to claim 1, characterized in that: The method of obtaining a similarity score between the initial embedding representation of the travel mode and the initial embedding representation of the user object and its features of the current travel includes: constructing an interaction matrix M2∈R between the historical user objects and the travel modes according to the big travel data X×Y , the interaction matrix between user objects and travel modes is used to define the second adjacency matrix that represents the graph relationship between users and travel modes: Then the user objects and travel modes are embedded and propagated iteratively through the second adjacency matrix A2: Among them, D2 is the degree matrix of the adjacency matrix A2, which is in the form of a diagonal matrix. is the embedding representation matrix composed of the embedding representation of the user object and its features and the embedding of the travel mode, and k is the number of iterations; Aggregate all the embedding propagation results. In the specific implementation process, this application adopts weighted aggregation: E U,C =γ0E 0 +γ1E 1 +...+c K E K ; Among them, γ0,γ1,…,γ J To add weights to the weighted aggregation process, the final user embedding matrix E is decomposed from the aggregation result. U and the travel mode embedding matrix E C ; The similarity score is calculated using the inner product of the final user and travel mode embedding representations.

7. The travel route planning method based on the recommendation principle according to claim 1, characterized in that: When training the model, the loss function used for training is as follows: For a given positive sample, sample a set of negative samples R n ,V n To minimize the probability of negative samples, C(r p ,v p ) is the positive route r preferred by user u p and the positive vehicle travel feature v preferred by user u p is the combination of user u’s preferred travel mode, s(,) is the similarity function for the corresponding object embedding representation, C(r p ,v p ) is the negative sub-route r abandoned by user u n and the negative vehicle travel characteristics v abandoned by user u n is the combination of , that is, the travel mode item abandoned by user u, and α is the set hyper parameter.

8. The travel route planning method based on the recommendation principle according to claim 7, characterized in that: The meaning of loss function: make but The loss function is The upper bound of .

9. A travel route planning device based on recommendation principle, characterized in that: include: At least one processing unit, the processing unit is connected to a storage unit via a bus unit, the storage unit stores a computer program, and when the computer program is executed by the processing unit, the travel path planning method based on the recommendation principle as described in any of claims 1-8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the travel route planning method based on the recommendation principle as described in any of claims 1 to 8 is implemented.

Images