Travel distance prediction method and device, server, medium and product

Abstract

The invention relates to a mileage prediction method and device, a server, a medium and a product, and relates to the technical field of automobiles, the method comprises the following steps: obtaining initial travel data and historical travel data of a current vehicle; extracting historical behavior characteristics of the user in the historical travel data, and identifying discrete characteristics and continuous characteristics in the initial travel data and the historical behavior characteristics; the discrete features and the continuous features are input into a prediction model, the prediction model outputs the predicted driving mileage of the current driving of the vehicle, the prediction model comprises a deep neural network, a decomposition network and a fusion network, the fusion network generates a matrix according to the discrete features and the continuous features, the matrix is input into the deep neural network and the decomposition network, and the prediction model outputs the predicted driving mileage of the current driving of the vehicle. And the fusion model splices respective outputs of the deep neural network and the decomposition network to obtain the predicted driving mileage of the current driving of the current vehicle. Therefore, the problems of poor accuracy of prediction of the driving mileage of the vehicle, poor user experience and the like in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of automobile technology, and in particular to a mileage prediction method, device, server, medium and product. Background Art

[0002] The battery is the core of electric vehicles. Adjusting the battery to operate in a suitable state will help extend battery life and improve user experience. For example, during vehicle driving, the battery control system will monitor the battery temperature in real time and perform thermal management on the battery: heating when the actual temperature is lower than the applicable temperature, and cooling when the actual temperature is higher than the applicable temperature. In actual scenarios, the distance and behavior of each user's trip are different, so the degree of battery use is also different. However, the battery control system function is relatively rigid, and it is impossible to optimize the control of battery thermal management in combination with the user's driving habits.

[0003] In view of the above status quo and needs, the following findings were found through the search of existing public literature:

[0004] In the related art, Publication No. (CN115742865A) proposes a thermal management method and system for electric vehicles based on mileage prediction. The method uses a training set to train a machine learning model to obtain a trained machine learning model, obtains real-time data related to the user's driving habits, and uses the trained machine learning model to predict the mileage and time that the user will travel, and performs energy-saving thermal management on the battery according to the mileage and time that the user will travel; the system includes a prediction module, a control module, and an energy-saving thermal management module. However, the scheme achieves prediction by introducing basic mileage, power, speed, etc., which cannot reflect the individual behavior differences and group behavior differences of users, thereby reducing the adaptability of individual differences of users, thereby reducing the accuracy of mileage prediction, and reducing the applicability of mileage prediction to individual differences of users, resulting in poor optimization efficiency of battery thermal management, reduced user experience, reduced user driving safety, and other issues.

[0005] Publication No. (CN114626571A) proposes a vehicle mileage prediction method and system, which obtains the original driving data of the owner; if the driving time in the original driving data is greater than or equal to the predetermined time period, the driving data corresponding to the predetermined time period in the original travel data is obtained; after preprocessing the driving data corresponding to the predetermined time period in the original travel data, the target driving data is generated; the target driving data is classified by a hierarchical clustering algorithm to generate a vehicle owner classification; according to the classified vehicle owners, corresponding prediction models are created respectively, and the vehicle owner's annual mileage prediction results are obtained according to the prediction model. However, this solution only further realizes the prediction according to similar user groups, and does not reflect the individual behavior differences of users, thereby reducing the accuracy of mileage prediction and the applicability of mileage prediction to individual differences of users, resulting in poor optimization efficiency of battery thermal management, reduced user experience, and reduced user driving safety. Summary of the invention

[0006] One of the objects of the present invention is to provide a mileage prediction method to solve the problems of poor accuracy of vehicle mileage prediction and poor user experience in the related art; the second object is to provide another mileage prediction method; the third object is to provide a mileage prediction device; the fourth object is to provide another mileage prediction device; the fifth object is to provide a server; the sixth object is to provide a computer-readable storage medium; the seventh object is to provide a computer program product.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] A mileage prediction method is applied to a server, and includes the following steps: obtaining the starting trip data and historical trip data of a current vehicle; extracting the historical behavior features of users in the historical trip data, and identifying the discrete features and continuous features in the starting trip data and the historical behavior features; inputting the discrete features and the continuous features into a prediction model, and the prediction model outputs the predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network and a fusion network, the fusion network generates a matrix according to the discrete features and the continuous features, and the matrix is ​​respectively input into the deep neural network and the decomposition network, and the fusion model splices the outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

[0009] According to the above technical means, the embodiments of the present invention predict the vehicle's mileage based on big data according to the individual behavior differences and group behavior differences of users, and realize real-time prediction of the vehicle's mileage through the deep integration of the big data algorithm model and the power battery thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual user differences, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0010] Furthermore, before the starting travel data and historical behavior characteristics are input into the prediction model, it also includes: obtaining historical driving data of multiple vehicles; extracting charging and discharging travel data of each vehicle in the historical driving data; establishing a travel factor database and a historical behavior database based on the charging and discharging travel data of each vehicle; generating a training data set based on the travel factor database and the historical behavior database, and using the training data set to train the prediction model.

[0011] According to the above technical means, the embodiment of the present invention establishes a travel factor database and a historical behavior database by establishing charging and discharging travel data of multiple vehicles, thereby incorporating a wider range of data samples, which helps to capture more comprehensive driving behavior patterns, thereby improving the generalization ability of the prediction model and its adaptability to individual differences of users.

[0012] Furthermore, the travel factor database includes at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature, wherein the starting travel factor includes at least one of a discharge starting power, a discharge starting time, and a discharge starting mileage, and at least one historical behavior feature includes at least one of the last discharge mileage, the number of discharge trips within a first preset time period, the time interval to the last discharge trip, the box line indicator of the discharge trip within the first preset time period, the time interval to the last charge, the last charge starting power, and the last charge starting time.

[0013] According to the above technical means, the embodiment of the present invention deeply analyzes the user's driving habits and battery usage patterns, so that the prediction model can take into account the user's daily travel patterns, analyze the number of discharge trips, box line indicators, etc. in the historical behavior database, and predict the user's driving tendency and battery usage intensity under specific conditions. Combined with the time interval from the last charge or discharge, the prediction model can determine the current charging state of the battery and the expected discharge demand, prevent over-discharge or heat the battery in advance when necessary to reduce energy loss, thereby improving the personalization level and accuracy of mileage prediction and optimizing the battery thermal management control policy.

[0014] Furthermore, a training data set is generated according to the travel factor database and the historical behavior database, including: extracting the starting travel factor and historical behavior characteristics of any vehicle; and concatenating the starting travel factor and the historical behavior characteristics to obtain the training data set.

[0015] According to the above technical means, the embodiment of the present invention obtains a training data set by splicing the starting trip factor and the historical behavior characteristics, ensuring that the training data set contains detailed information related to the specific vehicle and user behavior pattern. This personalized data integration helps the model to more accurately understand the unique driving habits and preferences of each vehicle and user. This improves the universal applicability and accuracy of mileage prediction.

[0016] Furthermore, the prediction model is trained using the training data set, including: identifying discrete features and continuous features in the training data set; inputting the discrete features and continuous features into the prediction model, and the prediction model outputs the predicted mileage; calculating the mileage error between the predicted mileage and the actual mileage, optimizing the prediction model according to the mileage error, and stopping the training until the mileage error is less than or equal to the preset error.

[0017] According to the above technical means, the embodiments of the present invention realize the generalization of high-dimensional and low-dimensional cross information, adapt to the situation where discrete information and continuous information coexist in the user's real driving scenes, expand the applicable scenarios of the prediction model, and enable it to effectively cope with complex and changeable real driving scenes; by comparing the predicted mileage and the actual mileage to determine whether to stop the training of the prediction model, the accuracy of the trained prediction model is improved.

[0018] Furthermore, after the prediction model outputs the predicted mileage of the current vehicle for this trip, it also includes: generating a thermal management control strategy for the current vehicle for this trip according to the predicted mileage.

[0019] According to the above-mentioned technical means, the embodiment of the present invention can input the travel factors and historical behavior characteristics as training data into the DeepFM deep learning model for training based on the RFM model and deep learning technology, so as to realize personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and then realize real-time dynamic adjustment of the thermal management control strategy.

[0020] Furthermore, the thermal management control strategy includes: if the predicted mileage is less than or equal to the mileage threshold, the power battery is not heated; if the predicted mileage is greater than the mileage threshold, a heating request for the power battery is determined according to the current vehicle driving power demand.

[0021] According to the above technical means, the embodiment of the present invention determines whether to heat the power battery by comparing the predicted mileage with the mileage threshold, thereby improving the accuracy of the power battery thermal management control policy.

[0022] Furthermore, a heating request for the power battery is determined according to the current driving power demand of the vehicle, including: if the maximum discharge power of the power battery is less than the driving power demand, requesting the power battery to be heated; if the maximum discharge power of the power battery is greater than or equal to the driving power demand, not heating the power battery.

[0023] According to the above technical means, the embodiment of the present invention improves the accuracy of real-time adjustment of the power battery thermal management control policy by comparing the maximum discharge power of the power battery with the driving power demand.

[0024] Furthermore, before determining the heating request of the power battery according to the current driving power demand of the vehicle, it also includes: obtaining the travel battery power distribution of the current vehicle within a second preset time period; and predicting the driving power demand of the current vehicle in multiple stages according to the travel battery power distribution.

[0025] According to the above technical means, the embodiment of the present invention predicts the driving power demand of the current vehicle in multiple stages through the trip battery power distribution, which can more accurately understand the actual power consumption of the vehicle under different conditions and provide more refined data support for the thermal management strategy.

[0026] A mileage prediction method is applied to a server, wherein the method comprises the following steps: obtaining historical driving data of a plurality of vehicles; extracting charging and discharging travel data of each vehicle in the historical driving data; establishing a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle, wherein the travel factor database comprises at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database comprises at least one historical behavior feature; generating a training data set according to the travel factor database and the historical behavior database, training a prediction model using the training data set, and determining the predicted mileage of the current vehicle for this trip based on the trained prediction model.

[0027] According to the above-mentioned technical means, the vehicle's mileage is predicted based on big data according to the individual behavior differences and group behavior differences of users, and the real-time prediction of the vehicle's mileage is achieved through the deep integration of the big data algorithm model and the power battery thermal management control strategy; through the RFM model and deep learning technology, the travel factor and historical behavior characteristics are input as training data into the DeepFM deep learning model for training, to achieve personalized travel prediction for each user, that is, "thousands of faces" mileage prediction, and then to achieve real-time dynamic adjustment of the thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0028] A mileage prediction device is applied to a server, and includes: a first acquisition module, used to obtain the starting trip data and historical trip data of a current vehicle; a first extraction module, used to extract the historical behavior characteristics of the user in the historical trip data, and identify the discrete characteristics and continuous characteristics in the starting trip data and the historical behavior characteristics; an input module, used to input the discrete characteristics and the continuous characteristics into a prediction model, and the prediction model outputs the predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network and a fusion network, the fusion network generates a matrix according to the starting trip data and the historical behavior characteristics, and the matrix is ​​input into the deep neural network and the decomposition network respectively, and the fusion model splices the outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

[0029] Furthermore, before the discrete features and the continuous features are input into the prediction model, the input module is also used to obtain the historical driving data of multiple vehicles; extract the charging and discharging travel data of each vehicle in the historical driving data; establish a travel factor database and a historical behavior database based on the charging and discharging travel data of each vehicle; generate a training data set based on the travel factor database and the historical behavior database, and use the training data set to train the prediction model.

[0030] Furthermore, the travel factor database includes at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature, wherein the starting travel factor includes at least one of a discharge starting power, a discharge starting time, and a discharge starting mileage, and at least one historical behavior feature includes at least one of the last discharge mileage, the number of discharge trips within a first preset time period, the time interval to the last discharge trip, the box line indicator of the discharge trip within the first preset time period, the time interval to the last charge, the last charge starting power, and the last charge starting time.

[0031] Furthermore, the input module is further used to extract the starting travel factor and historical behavior characteristics of any vehicle; and concatenate the starting travel factor and historical behavior characteristics to obtain a training data set.

[0032] Furthermore, the input module is further used to identify discrete features and continuous features in the training data set; input the discrete features and continuous features into the prediction model, and the prediction model outputs the predicted mileage; calculate the mileage error between the predicted mileage and the actual mileage, and optimize the prediction model according to the mileage error until the mileage error is less than or equal to the preset error, and stop training.

[0033] Furthermore, the first generating module is used to generate a thermal management control strategy for the current vehicle's current travel according to the predicted mileage after the prediction model outputs the predicted mileage of the current vehicle's current travel.

[0034] Furthermore, the thermal management control strategy includes: when the predicted mileage is less than or equal to the mileage threshold, not heating the power battery; when the predicted mileage is greater than the mileage threshold, determining a heating request for the power battery based on the current vehicle driving power demand.

[0035] Furthermore, the first generating module is further used to request heating of the power battery when the maximum discharge power of the power battery is less than the driving power requirement; and not to heat the power battery when the maximum discharge power of the power battery is greater than or equal to the driving power requirement.

[0036] Furthermore, before determining the power battery heating request according to the current vehicle's driving power demand, the first generating module is also used to obtain the current vehicle's travel battery power distribution within a second preset time period; and predict the current vehicle's driving power demand in multiple stages according to the travel battery power distribution.

[0037] A mileage prediction device is applied to a server, and includes: a second acquisition module for acquiring historical driving data of multiple vehicles; a second extraction module for extracting charging and discharging travel data of each vehicle in the historical driving data; an establishment module for establishing a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle, the travel factor database including at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database including at least one historical behavior feature; a second generation module for generating a training data set according to the travel factor database and the historical behavior database, training a prediction model using the training data set, and determining the predicted mileage of the current vehicle's current trip based on the trained prediction model.

[0038] A server includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the mileage prediction method as described in the above embodiment.

[0039] A computer-readable storage medium stores a computer program or instruction, which is executed by a processor to implement the mileage prediction method as described in the above embodiment.

[0040] A computer program product includes a computer program or instructions, and when the computer program or instructions are executed, the mileage prediction method as described in the above embodiment is implemented.

[0041] Beneficial effects of the present invention:

[0042] (1) The embodiments of the present invention predict the mileage of a vehicle based on big data according to the individual behavior differences and group behavior differences of users. By deeply integrating the big data algorithm model with the power battery thermal management control strategy, real-time prediction of the vehicle mileage is achieved, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving the user experience.

[0043] (2) The embodiment of the present invention establishes a travel factor database and a historical behavior database by establishing the charging and discharging travel data of multiple vehicles, thereby incorporating a wider range of data samples, which helps to capture more comprehensive driving behavior patterns, thereby improving the generalization ability of the prediction model and its adaptability to individual differences of users.

[0044] (3) The embodiment of the present invention deeply analyzes the user's driving habits and battery usage patterns, so that the prediction model can take into account the user's daily travel patterns and analyze the number of discharge trips, box line indicators, etc. in the historical behavior database. It can predict the user's driving tendency and battery usage intensity under specific conditions. Combined with the time interval since the last charge or discharge, the prediction model can determine the current charging state of the battery and the expected discharge demand, prevent over-discharge or heat the battery in advance to reduce energy loss when necessary, thereby improving the personalization level and accuracy of the mileage prediction and optimizing the battery thermal management control policy.

[0045] (4) The embodiment of the present invention obtains a training data set by splicing the starting trip factor and the historical behavior characteristics, ensuring that the training data set contains detailed information related to the specific vehicle and user behavior pattern. This personalized data integration helps the model to more accurately understand the unique driving habits and preferences of each vehicle and user. This improves the universal applicability and accuracy of mileage prediction.

[0046] (5) The embodiments of the present invention realize the generalization of high-dimensional and low-dimensional cross-information, adapt to the situation where discrete information and continuous information coexist in the user's real driving scenes, expand the applicable scenarios of the prediction model, and enable it to effectively cope with complex and changeable real driving scenes; by comparing the predicted mileage and the actual mileage to determine whether to stop the training of the prediction model, the accuracy of the trained prediction model is improved.

[0047] (6) The embodiment of the present invention can input the travel factors and historical behavior characteristics as training data into the DeepFM deep learning model for training based on the RFM model and deep learning technology, thereby realizing personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and further realizing real-time dynamic adjustment of the thermal management control strategy.

[0048] (7) The embodiment of the present invention determines whether to heat the power battery by comparing the predicted mileage with the mileage threshold, thereby improving the accuracy of the power battery thermal management control policy.

[0049] (8) The embodiment of the present invention improves the accuracy of real-time adjustment of the power battery thermal management control policy by comparing the maximum discharge power of the power battery with the driving power demand.

[0050] (9) The embodiment of the present invention predicts the driving power requirements of the current vehicle in multiple stages through the travel battery power distribution, which can more accurately understand the actual power consumption of the vehicle under different conditions and provide more refined data support for the thermal management strategy.

[0051] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a flow chart of a mileage prediction method according to the present invention;

[0053] Figure 2 A schematic diagram of a mileage prediction method provided according to an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of selecting common activity areas according to an embodiment of the present invention;

[0055] Figure 4 A schematic diagram of establishing a database according to an embodiment of the present invention;

[0056] Figure 5 A schematic diagram of a construction prediction model provided according to an embodiment of the present invention;

[0057] Figure 6 A schematic diagram of a thermal management control strategy provided according to an embodiment of the present invention;

[0058] Figure 7 A schematic diagram of a flow chart of another mileage prediction method according to the present invention;

[0059] Figure 8 A block diagram of a mileage prediction device according to the present invention;

[0060] Fig. 9 A block diagram of another mileage prediction device according to the present invention;

[0061] Fig.10 FIG. 1 is a schematic diagram of the structure of a server according to the present invention.

[0062] Among them: 701 - first acquisition module; 702 - first extraction module; 703 - input module; 801 - second acquisition module; 802 - second extraction module; 803 - establishment module; 804 - second generation module; 901 - memory; 902 - processor and 903 - communication interface. DETAILED DESCRIPTION

[0063] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0064] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0065] The present invention provides a mileage prediction method, device, server, medium and product. The mileage prediction method is described below. Figure 1 It is a flowchart of a mileage prediction method provided by an embodiment of the present invention.

[0066] like Figure 1 As shown, the mileage prediction method is applied to a server, which includes the following steps:

[0067] In step S101 , the initial travel data and historical travel data of the current vehicle are obtained.

[0068] It should be noted that the server may be a cloud platform, the type of which is not specifically limited herein, and those skilled in the art may set it according to actual conditions.

[0069] Among them, Figure 2 As shown, the starting trip data and historical trip data are stored on the cloud platform, and the longitude and latitude of the starting and ending points of the trips are clustered to determine the user's common activity areas.

[0070] Specifically, if Figure 3As shown, the starting trip data of n historical vehicles is read from the cloud platform, and multiple longitude and latitude points at the start and end of the trip are input into the dbscan (Density-Based Spatial Clustering of Applications with Noise) method. Multiple vehicle activity areas and corresponding activity times are clustered, and the three categories with the largest number of longitude and latitude points are selected as common activity areas.

[0071] Among them, the dbscan method is a density-based clustering method that can identify areas with sufficiently high density as clusters and can find clusters of arbitrary shapes in a noisy spatial database.

[0072] It should be noted that, in the embodiment of the present invention, both the starting travel data and the historical travel data of the current vehicle can be obtained through the vehicle's sensors, wherein the type of the sensor is not specifically limited herein.

[0073] In step S102, historical behavior features of the user in the historical travel data are extracted, and discrete features and continuous features in the starting travel data and the historical behavior features are identified.

[0074] It can be understood that in the embodiment of the present invention, the discrete features can be features of the starting trip data and the historical behavior features in discrete form; the continuous features can be features of the starting trip data and the historical behavior features in continuous form.

[0075] Specifically, if Figure 4 As shown, the embodiment of the present invention can extract the historical behavior characteristics of users in historical travel data based on the concept of the RFM model, thereby constructing a user behavior value factor and forming an RFM model.

[0076] Among them, the RFM model can include three core indicators: recency, frequency and strength. The RFM model segments and evaluates the user's historical behavior characteristics based on these three core indicators to accurately predict mileage and optimize the battery's thermal management strategy.

[0077] It is understandable that the extracted user behavior value factors can be stored in the cloud platform.

[0078] It should be noted that the detailed steps for constructing the user behavior value factor based on the RFM concept can be found in the following embodiment and will not be repeated here.

[0079] Therefore, in the embodiment of the present invention, by designing a method for constructing a user behavior value factor to explore the characteristics of the user's driving behavior, it is possible to better adapt to individual differences among users and improve prediction accuracy.

[0080] In step S103, discrete features and continuous features are input into a prediction model, and the prediction model outputs a predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network, and a fusion network. The fusion network generates a matrix according to the discrete features and the continuous features, and the matrix is ​​respectively input into the deep neural network and the decomposition network. The fusion model splices the outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

[0081] It can be understood that the prediction model can be a deep learning DeepFM model; the fusion network generates a matrix based on the starting trip data and historical behavior characteristics, taking the x matrix as an example; the fusion network takes the sigmoid algorithm as an example; the deep neural network takes the DNN network as an example; the decomposition network takes the FM decomposition machine as an example.

[0082] Specifically, establish a mileage prediction model: initialize the deep learning DeepFM model. The specific steps are as follows: Figure 5 As shown:

[0083] (1) Construct a 2-layer fully connected DNN network: input x, perform linear(x) and relu(x) processing in each fully connected layer, and output the processed x;

[0084] (2) Construct a second-order cross FM decomposition machine: input x, refer to the embedding dimension emb_num and the total dimension fea_num, define the global bias matrix w0, the first-order weight matrix w1, and the second-order cross matrix w2, and calculate the first-order cross (w0+x×w1) and the second-order cross ((x×w2)^2-x^2×w2^2), and output the sum of the two;

[0085] (3) Construct the DeepFM model: input x, embed the discrete features into a matrix of dimension emb_num, and concatenate it with the continuous feature matrix to generate a matrix x of total dimension fea_num; input the x matrix into the (1) and (2) models respectively, concatenate the two outputs, perform sigmoid calculation, and obtain the output result output.

[0086] In an embodiment of the present invention, before inputting discrete features and continuous features into a prediction model, it also includes: acquiring historical driving data of multiple vehicles; extracting charging and discharging travel data of each vehicle in the historical driving data; establishing a travel factor database and a historical behavior database based on the charging and discharging travel data of each vehicle; generating a training data set based on the travel factor database and the historical behavior database, and using the training data set to train the prediction model.

[0087] It should be noted that, in the embodiment of the present invention, the historical behavior database is a user behavior value factor database.

[0088] Among them, the travel factor database includes at least one of the starting travel factor, the ongoing travel factor, whether the vehicle is located in a common activity area and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature, wherein the starting travel factor includes at least one of the discharge starting power, the discharge starting time and the discharge starting mileage, and at least one historical behavior feature includes at least one of the last discharge mileage, the number of discharge trips within a first preset time period, the time interval to the last discharge trip, the box line indicator of the discharge trip within the first preset time period, the time interval to the last charge, the last charge starting power and the last charge starting time.

[0089] It can be understood that, in the embodiment of the present invention, the first preset duration is 7 days as an example.

[0090] Specifically, if Figure 4 As shown in Table 1, for each charging and discharging trip of n vehicles, the calculated factors include: starting power, starting time, starting mileage, trip remaining power, remaining time, remaining mileage, whether it is in a common activity area, and whether it is in an activity time. The above data are stored in the trip factor database.

[0091] Table 1

[0092] Trip Number Starting day of the week Starting power …… Whether to start from the activity point Mileage 1 6 79 …… 1 50 2 3 30 …… 0 90

[0093] On the basis of the trip factor database, for each discharge trip that needs to predict the mileage, based on the RFM model, the "time since the last trip" is designed as the proximate cause, the "frequency of trips in the last 7 days" is designed as the frequency cause, and the "mileage, power consumption, and duration of the trip" are designed as the strong cause. The factors for the calculation of the discharge status include: the time since the last discharge trip, the frequency of discharge trips in the last 7 days, the mileage, power consumption, and duration of the last discharge trip, and the box line indicators of the mileage, power consumption, and duration of discharge trips in the last 7 days, as shown in Table 2; the factors for the calculation of the charging status are similar, but do not include the "mileage" indicator, and the "power consumption" indicator is replaced by the "charging" indicator. The above data are all stored in the user behavior value factor database.

[0094] Table 2

[0095]

[0096] Therefore, in an embodiment of the present invention, a multi-dimensional, multi-state user behavior value factor database representing the user's historical behavior characteristics can be constructed based on the RFM model, which includes factors for charging state calculation and factors for discharging state, so as to better adapt to the individual differences in user driving behavior and improve the accuracy of mileage prediction according to the battery's charging and discharging state.

[0097] In an embodiment of the present invention, a training data set is generated based on a travel factor database and a historical behavior database, including: extracting a starting travel factor and historical behavior features of any vehicle; and concatenating the starting travel factor and historical behavior features to obtain a training data set.

[0098] Specifically, vehicles with historical discharge trips are selected, and multiple discharge trip data and charging trip data are extracted for them, wherein the starting trip factor and user behavior value factor of each discharge trip are spliced; the starting trip factor and user behavior value factor of each charging trip are spliced ​​as the training data set. Table 3 shows the spliced ​​starting trip factor and user behavior value factor of the discharge trip.

[0099] Table 3

[0100]

[0101] In an embodiment of the present invention, a prediction model is trained using a training data set, including: identifying discrete features and continuous features in the training data set; inputting the discrete features and continuous features into the prediction model, and the prediction model outputs a predicted mileage; calculating the mileage error between the predicted mileage and the actual mileage, optimizing the prediction model according to the mileage error, and stopping the training until the mileage error is less than or equal to a preset error.

[0102] It can be understood that in the embodiment of the present invention, discrete features are taken as sparse_inputs as an example; continuous features are taken as dense_inputs as an example; predicted mileage is taken as outputs as an example; actual mileage is taken as real_MLG as an example; mileage error is taken as MAE as an example; and the preset error is taken as 3 kilometers as an example.

[0103] Specifically, the discrete features sparse_inputs and continuous features dense_inputs in the training data set are distinguished, and then input into the prediction model DeepFM, the loss function of the predicted mileage outputs and the real mileage real_MLG is calculated, and gradient descent training is performed.

[0104] It should be noted that the type of loss function is not specifically limited here, and those skilled in the art can set it according to actual conditions.

[0105] Specifically, if Figure 2As shown in the figure, if the thermal management control strategy of the power battery believes that the error of the predicted mileage outputs within 3 kilometers will not affect its strategy adjustment, MAE is used as an evaluation indicator to test the model effect, with 3 kilometers as the threshold. The effect of the prediction model DeepFM is tested using an independent test set. If the MAE of the prediction model DeepFM on the independent test set is less than or equal to 3 kilometers, the model is saved. Otherwise, the samples and features are adjusted and the model is retrained.

[0106] Therefore, the embodiment of the present invention establishes a big data algorithm prediction model through a deep learning model, thereby generalizing high-dimensional and low-dimensional cross-information, adapting to the situation where discrete information and continuous information coexist in the user's real driving scenarios, expanding the applicable scenarios of the prediction model, and having practical significance.

[0107] In the embodiment of the present invention, after the prediction model outputs the predicted mileage of the current vehicle for this trip, it also includes: generating a thermal management control strategy for the current vehicle for this trip according to the predicted mileage.

[0108] Among them, the predicted mileage takes pred_MLG as an example.

[0109] Specifically, if Figure 6 As shown, when the vehicle starts driving, the model is called to predict the mileage, and the thermal management control strategy is adjusted and optimized. For vehicle A with a historical discharge journey, at the beginning of a new discharge journey, the power, mileage, activity area determination, activity time and other data are obtained from the vehicle side and the cloud platform at the same time, and the mileage pred_MLG of this trip is predicted in the prediction model saved by the cloud platform and sent to the vehicle side.

[0110] Therefore, the embodiment of the present invention can adjust and optimize the thermal management control strategy of the current vehicle in real time by predicting the mileage, thereby improving the accuracy and timeliness of the thermal management of the power battery.

[0111] Among them, the thermal management control strategy includes: if the predicted mileage is less than or equal to the mileage threshold, the power battery is not heated; if the predicted mileage is greater than the mileage threshold, the power battery heating request is determined according to the current vehicle driving power demand.

[0112] It should be noted that the mileage threshold is taken as thr_MLG.

[0113] As a possible way to achieve this, Figure 6 As shown, if pred_MLG<=thr_MLG, the vehicle is set to a short-range driving mode, in which no battery heating request is sent;

[0114] As another possible implementation, Figure 6As shown, if pred_MLG>thr_MLG, the vehicle is set to the remote driving mode, in which a battery heating request is sent.

[0115] Therefore, the embodiment of the present invention can determine whether the thermal management control strategy heats the battery by comparing the predicted mileage with the mileage threshold, thereby improving the accuracy of the thermal management control strategy.

[0116] In an embodiment of the present invention, before determining a heating request for the power battery according to the current vehicle's driving power demand, it also includes: obtaining a travel battery power distribution of the current vehicle within a second preset time period; and predicting the current vehicle's driving power demand in multiple stages according to the travel battery power distribution.

[0117] It should be noted that, in the embodiment of the present invention, the second preset time duration is 30 days as an example; the multiple stages may include a trip start stage, a trip mid-stage and a trip end stage; and the driving power demand is pred_Pi as an example.

[0118] In an embodiment of the present invention, a heating request for the power battery is determined according to the current driving power demand of the vehicle, including: if the maximum discharge power of the power battery is less than the driving power demand, requesting the power battery to be heated; if the maximum discharge power of the power battery is greater than or equal to the driving power demand, not heating the power battery.

[0119] Specifically, if Figure 6 As shown in the figure, in the long-distance driving mode, based on the battery power distribution of vehicle A in the similar mileage range in the past 30 days, the user's driving power demand pred_Pi at the beginning, middle and end of the journey is estimated respectively. In each stage, if the current maximum battery discharge power < driving power demand pred_Pi, battery heating is requested, otherwise battery heating is not requested.

[0120] Therefore, the embodiment of the present invention determines whether the thermal management control policy of the power battery heats the battery by comparing the maximum discharge power of the power battery with the driving power demand, thereby achieving energy conservation and emission reduction and reducing energy consumption.

[0121] The mileage prediction method of the embodiment of the present invention is described below with reference to a specific embodiment. Figure 2 As shown, the specific process method is as follows:

[0122] (1) Build a multi-dimensional and multi-state user behavior value factor database based on the RFM model;

[0123] (2) Train the big data algorithm model for predicting mileage, concatenate factors to obtain the training data set, input the model to reach the required test threshold, and save the model;

[0124] (3) Predict the mileage of vehicle A, calculate the user behavior value factor of vehicle A at the beginning of a new trip, and obtain the predicted mileage result after splicing other factors into the model to assist in adjusting the battery thermal management strategy.

[0125] A mileage prediction method proposed in an embodiment of the present invention predicts the mileage of a vehicle based on big data according to the individual behavior differences and group behavior differences of users, and realizes real-time prediction of the vehicle's mileage through deep integration of the big data algorithm model and the power battery thermal management control strategy; by integrating the RFM model and deep learning technology, the travel factor and the historical behavior characteristics are input as training data into the DeepFM deep learning model for training, so as to realize personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and then realize real-time dynamic adjustment of the thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0126] Figure 7 FIG. 1 is a flow chart of another mileage prediction method provided by an embodiment of the present invention. Figure 7 As shown, the mileage prediction method is applied to a server, which includes the following steps:

[0127] In step S601, historical driving data of multiple vehicles is obtained.

[0128] In step S602, the charging and discharging travel data of each vehicle in the historical driving data is extracted.

[0129] In step S603, a travel factor database and a historical behavior database are established based on the charging and discharging travel data of each vehicle. The travel factor database includes at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time. The historical behavior database includes at least one historical behavior feature.

[0130] In step S604, a training data set is generated according to the travel factor database and the historical behavior database, the prediction model is trained using the training data set, and the predicted mileage of the current vehicle for this trip is determined based on the trained prediction model.

[0131] It should be noted that the above explanations of the embodiment of the mileage prediction method are also applicable to the mileage prediction method of this embodiment, and will not be repeated here.

[0132] According to the mileage prediction method proposed in the embodiment of the present invention, the mileage of the vehicle is predicted based on big data according to the individual behavior differences and group behavior differences of users, and the real-time prediction of the vehicle's mileage is achieved through the deep integration of the big data algorithm model and the power battery thermal management control strategy; by integrating the RFM model and deep learning technology, the travel factor and the historical behavior characteristics are input as training data into the DeepFM deep learning model for training, so as to achieve personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and then realize real-time dynamic adjustment of the thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0133] Next, a mileage prediction device according to an embodiment of the present invention is described with reference to the accompanying drawings.

[0134] Figure 8 It is a block diagram of a mileage prediction device according to an embodiment of the present invention.

[0135] like Figure 8 As shown, the mileage prediction device is applied to a server, and includes: a first acquisition module 701 , a first extraction module 702 and an input module 703 .

[0136] Among them, the first acquisition module 701 is used to obtain the starting travel data and historical travel data of the current vehicle; the first extraction module 702 is used to extract the historical behavior characteristics of the user in the historical travel data, and identify the discrete features and continuous features in the starting travel data and the historical behavior characteristics; the input module 703 is used to input the discrete features and the continuous features into the prediction model, and the prediction model outputs the predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network and a fusion network, the fusion network generates a matrix according to the discrete features and the continuous features, and the matrix is ​​input into the deep neural network and the decomposition network respectively, and the fusion model splices the outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

[0137] In the embodiment of the present invention, before the discrete features and the continuous features are input into the prediction model, the input module 703 is also used to obtain the historical driving data of multiple vehicles; extract the charging and discharging travel data of each vehicle in the historical driving data; establish a travel factor database and a historical behavior database based on the charging and discharging travel data of each vehicle; generate a training data set based on the travel factor database and the historical behavior database, and use the training data set to train the prediction model.

[0138] Among them, the travel factor database includes at least one of the starting travel factor, the ongoing travel factor, whether the vehicle is located in a common activity area and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature, wherein the starting travel factor includes at least one of the discharge starting power, the discharge starting time and the discharge starting mileage, and at least one historical behavior feature includes at least one of the last discharge mileage, the number of discharge trips within a first preset time period, the time interval to the last discharge trip, the box line indicator of the discharge trip within the first preset time period, the time interval to the last charge, the last charge starting power and the last charge starting time.

[0139] In the embodiment of the present invention, the input module 703 is further used to extract the starting travel factor and historical behavior characteristics of any vehicle; and concatenate the starting travel factor and historical behavior characteristics to obtain a training data set.

[0140] In an embodiment of the present invention, the input module 703 is further used to identify discrete features and continuous features in a training data set; input the discrete features and continuous features into a prediction model, and the prediction model outputs a predicted mileage; calculates the mileage error between the predicted mileage and the actual mileage, and optimizes the prediction model according to the mileage error, and stops training until the mileage error is less than or equal to a preset error.

[0141] In an embodiment of the present invention, the first generating module is used to generate a thermal management control strategy for the current vehicle's current travel according to the predicted mileage after the prediction model outputs the predicted mileage of the current vehicle's current travel.

[0142] Among them, the thermal management control strategy includes: when the predicted mileage is less than or equal to the mileage threshold, the power battery is not heated; when the predicted mileage is greater than the mileage threshold, the power battery heating request is determined according to the current vehicle driving power demand.

[0143] In an embodiment of the present invention, the first generating module is further used to request heating of the power battery when the maximum discharge power of the power battery is less than the driving power requirement; and not to heat the power battery when the maximum discharge power of the power battery is greater than or equal to the driving power requirement.

[0144] In the embodiment of the present invention, before determining the heating request of the power battery according to the driving power demand of the current vehicle, the first generating module is further used to obtain the travel battery power distribution of the current vehicle within the second preset time length; predict the driving power demand of the current vehicle in multiple stages according to the travel battery power distribution

[0145] It should be noted that the above explanations of the mileage prediction method embodiment are also applicable to the mileage prediction device of this embodiment, and will not be repeated here.

[0146] The mileage prediction device proposed in the embodiment of the present invention predicts the mileage of the vehicle based on big data according to the individual behavior differences and group behavior differences of users, and realizes real-time prediction of the vehicle's mileage through the deep integration of the big data algorithm model and the power battery thermal management control strategy; by integrating the RFM model and deep learning technology, the travel factor and the historical behavior characteristics are input as training data into the DeepFM deep learning model for training, so as to realize personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and then realize real-time dynamic adjustment of the thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0147] Next, another mileage prediction device according to an embodiment of the present invention will be described with reference to the accompanying drawings.

[0148] Fig. 9 It is a block diagram of a mileage prediction device according to an embodiment of the present invention.

[0149] like Fig. 9 As shown, the driving mileage prediction device is applied to a server, and includes: a second acquisition module 801 , a second extraction module 802 , a creation module 803 and a second generation module 804 .

[0150] Among them, the second acquisition module 801 is used to obtain historical driving data of multiple vehicles; the second extraction module 802 is used to extract the charging and discharging travel data of each vehicle in the historical driving data; the establishment module 803 is used to establish a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle, the travel factor database includes at least one of the starting travel factor, the ongoing travel factor, whether the vehicle is located in a common activity area and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature; the second generation module 804 is used to generate a training data set according to the travel factor database and the historical behavior database, use the training data set to train the prediction model, and determine the predicted mileage of the current vehicle for this trip based on the trained prediction model.

[0151] It should be noted that the above explanations of the mileage prediction method embodiment are also applicable to the mileage prediction device of this embodiment, and will not be repeated here.

[0152] The mileage prediction device proposed in the embodiment of the present invention predicts the mileage of the vehicle based on the individual behavior differences and group behavior differences of users based on big data, and realizes real-time prediction of the vehicle's mileage through the deep integration of the big data algorithm model and the power battery thermal management control strategy; by integrating the RFM model and deep learning technology, the travel factor and the historical behavior characteristics are input as training data into the DeepFM deep learning model for training, so as to realize personalized travel prediction for each user, that is, "one thousand faces" mileage prediction, and then realize real-time dynamic adjustment of the thermal management control strategy, thereby improving the accuracy of mileage prediction, improving the applicability of mileage prediction to individual differences of users, improving the optimization efficiency of battery thermal management, reducing energy consumption, and improving user experience.

[0153] Fig.10 A schematic diagram of the structure of a server provided in an embodiment of the present invention. The server may include:

[0154] A memory 901 , a processor 902 , and a computer program stored in the memory 901 and executable on the processor 902 .

[0155] When the processor 902 executes the program, the mileage prediction method provided in the above embodiment is implemented.

[0156] Furthermore, the server also includes:

[0157] The communication interface 903 is used for communication between the memory 901 and the processor 902 .

[0158] The memory 901 is used to store computer programs that can be executed on the processor 902 .

[0159] The memory 901 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0160] If the memory 901, the processor 902 and the communication interface 903 are implemented independently, the communication interface 903, the memory 901 and the processor 902 can be connected to each other through a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.10Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0161] Optionally, in a specific implementation, if the memory 901, the processor 902 and the communication interface 903 are integrated on a chip, the memory 901, the processor 902 and the communication interface 903 can communicate with each other through an internal interface.

[0162] The processor 902 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention.

[0163] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program or instruction is stored. The computer program or instruction is executed by a processor to implement the above mileage prediction method.

[0164] An embodiment of the present invention further provides a computer program product, including a computer program or instructions, which, when executed, implements the mileage prediction method as described in the above embodiment.

[0165] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0166] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0167] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0168] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array, a field programmable gate array, etc.

[0169] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0170] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A mileage prediction method, characterized in that: The method is applied to a server and includes the following steps: Get the starting trip data and historical trip data of the current vehicle; Extracting historical behavior features of the user from the historical travel data, and identifying discrete features and continuous features in the starting travel data and the historical behavior features; The discrete features and the continuous features are input into a prediction model, and the prediction model outputs the predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network, and a fusion network, and the fusion network generates a matrix according to the discrete features and the continuous features, and the matrix is ​​respectively input into the deep neural network and the decomposition network, and the fusion model splices the respective outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

2. The mileage prediction method according to claim 1, characterized in that: Before the discrete features and the continuous features are input into the prediction model, the method further includes: Obtain historical driving data of multiple vehicles; Extracting charging and discharging travel data of each vehicle in the historical driving data; Establishing a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle; A training data set is generated according to the travel factor database and the historical behavior database, and the prediction model is trained using the training data set.

3. The mileage prediction method according to claim 2, characterized in that: The travel factor database includes at least one of a starting travel factor, an ongoing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time; the historical behavior database includes at least one historical behavior feature, wherein the starting travel factor includes at least one of a discharge starting power, a discharge starting time, and a discharge starting mileage; the at least one historical behavior feature includes at least one of the last discharge mileage, the number of discharge trips within a first preset time, the time interval to the last discharge trip, the box line index of the discharge trip within the first preset time, the time interval to the last charge, the last charge starting power, and the last charge starting time.

4. The mileage prediction method according to claim 3, characterized in that: The generating of a training data set according to the travel factor database and the historical behavior database comprises: Extract the starting trip factor and historical behavior characteristics of any vehicle; The training data set is obtained by concatenating the starting travel factor and the historical behavior feature.

5. The mileage prediction method according to claim 2, characterized in that: Training the prediction model using the training data set includes: Identifying discrete features and continuous features in the training data set; Inputting the discrete features and the continuous features into the prediction model, the prediction model outputting the predicted mileage; The mileage error between the predicted mileage and the actual mileage is calculated, and the prediction model is optimized according to the mileage error, and the training is stopped until the mileage error is less than or equal to a preset error.

6. The mileage prediction method according to claim 1, characterized in that: After the prediction model outputs the predicted mileage of the current vehicle, the method further includes: A thermal management control strategy for the current vehicle's current travel is generated according to the predicted mileage.

7. The mileage prediction method according to claim 6, characterized in that: The thermal management control strategy includes: If the predicted mileage is less than or equal to the mileage threshold, the power battery is not heated; If the predicted mileage is greater than the mileage threshold, a heating request for the power battery is determined according to the current driving power demand of the vehicle.

8. The mileage prediction method according to claim 7, characterized in that: The step of determining the heating request for the power battery according to the current driving power demand of the vehicle includes: If the maximum discharge power of the power battery is less than the driving power requirement, requesting power battery heating; If the maximum discharge power of the power battery is greater than or equal to the driving power requirement, the power battery is not heated.

9. The mileage prediction method according to claim 7, characterized in that: Before determining the heating request of the power battery according to the current driving power demand of the vehicle, the method further includes: Obtaining a battery power distribution of the current vehicle within a second preset time period; The driving power requirements of the current vehicle in multiple stages are predicted according to the travel battery power distribution.

10. A mileage prediction method, characterized in that: The method is applied to a server, wherein the method comprises the following steps: Obtain historical driving data of multiple vehicles; Extracting charging and discharging travel data of each vehicle in the historical driving data; Establishing a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle, wherein the travel factor database includes at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature; A training data set is generated according to the travel factor database and the historical behavior database, the prediction model is trained using the training data set, and the predicted mileage of the current vehicle for this trip is determined based on the trained prediction model.

11. A mileage prediction device, characterized in that: The device is applied to a server, wherein the device comprises: A first acquisition module (701) is used to acquire the starting travel data and historical travel data of the current vehicle; A first extraction module (702) is used to extract the historical behavior features of the user from the historical travel data, and identify discrete features and continuous features in the starting travel data and the historical behavior features; The input module (703) is used to input the discrete features and the continuous features into a prediction model, and the prediction model outputs the predicted mileage of the current vehicle for this trip, wherein the prediction model includes a deep neural network, a decomposition network and a fusion network, the fusion network generates a matrix according to the discrete features and the continuous features, and the matrix is ​​input into the deep neural network and the decomposition network respectively, and the fusion model splices the outputs of the deep neural network and the decomposition network to obtain the predicted mileage of the current vehicle for this trip.

12. A mileage prediction device, characterized in that: The device is applied to a server, wherein the device comprises: A second acquisition module (801) is used to acquire historical driving data of multiple vehicles; A second extraction module (802) is used to extract the charging and discharging travel data of each vehicle in the historical driving data; An establishing module (803) is used to establish a travel factor database and a historical behavior database according to the charging and discharging travel data of each vehicle, wherein the travel factor database includes at least one of a starting travel factor, a continuing travel factor, whether the vehicle is located in a common activity area, and whether the vehicle is in an activity time, and the historical behavior database includes at least one historical behavior feature; The second generation module (804) is used to generate a training data set based on the travel factor database and the historical behavior database, use the training data set to train the prediction model, and determine the predicted mileage of the current vehicle's current trip based on the trained prediction model.

13. A server, characterized in that: include: A memory (901), a processor (902), and a computer program stored in the memory (901) and executable on the processor (902), wherein the processor (902) executes the program to implement the mileage prediction method according to any one of claims 1 to 10.

14. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: The computer program or instruction is executed by a processor to implement the mileage prediction method according to any one of claims 1 to 10.

15. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed, the mileage prediction method according to any one of claims 1 to 10 is implemented.

Images