Vehicle function recommendation method, electronic equipment and computer storage medium

Abstract

The invention provides a vehicle function recommendation method, electronic equipment and a computer storage medium. The method comprises the steps that the current vehicle state of a vehicle is obtained; receiving first use data when the vehicle is in a vehicle state in a historical period; obtaining second use data according to the first use data, wherein the second use data represents data that each target function is preferentially used by the user in a historical period; obtaining third use data according to the first use data, wherein the third use data represents predicted data that each target function is preferentially used by the user; calculating a use probability value of each target function according to the second use data and the third use data; determining a target probability value greater than a preset probability threshold in the plurality of use probability values; and recommending a target function matched with the target probability value. According to the method and the device, the target function conforming to the user preference can be accurately recommended according to the historical use preference of the target function in the historical period.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a vehicle function recommendation method, electronic equipment and computer storage medium. Background Art

[0002] With the continuous development of social economy and vehicle technology, more and more families are buying cars. Compared with public transportation, cars can directly reach the destination from the departure point, saving waiting and transfer time, improving travel efficiency, and providing convenience and comfort for people's life and work.

[0003] At present, vehicles are equipped with various target functions, and users can use different target functions to meet their needs for entertainment, relaxation, etc. For example, users can have a seat massage in the vehicle, turn on the ambient light when parking, listen to music when parking, turn on the air conditioner before getting in the car, etc. However, there is a problem that it is cumbersome for users to turn on some target functions, resulting in a poor user experience. Summary of the invention

[0004] In view of this, the present application provides a vehicle function recommendation method, an electronic device and a computer storage medium, which can solve the problem that it is cumbersome for users to enable certain target functions, resulting in a poor user experience.

[0005] An embodiment of the present application provides a vehicle function recommendation method, which includes: obtaining a current vehicle state of a vehicle; receiving first usage data of the vehicle when it is in the vehicle state during a historical period, the first usage data being the number of times each target function of the vehicle is used during the historical period; obtaining second usage data based on the first usage data, the second usage data representing data on user preference for use of each target function during the historical period; obtaining third usage data based on the first usage data, the third usage data representing predicted data on user preference for use of each target function; calculating a usage probability value of each target function based on the second usage data and the third usage data; determining a target probability value that is greater than a preset probability threshold among a plurality of the usage probability values; and recommending the target function that matches the target probability value.

[0006] Compared with the related art, the embodiments of the present application have at least the following advantages: By obtaining the current vehicle state of the vehicle and receiving the first usage data when the vehicle was in the vehicle state during the historical period. From the first usage data, the second usage data of each target function preferred by the user during the historical period and the third usage data of each target function predicted to be preferred by the user can be obtained. Through the second usage data and the predicted third usage data, the usage probability value of each target function is calculated, and the target probability value greater than the preset probability threshold value among the multiple usage probability values ​​is determined. Therefore, based on the target probability value, the target function that meets the user's usage preference is accurately recommended to the user, so that the user can directly use the preferred target function without performing cumbersome operation steps, thereby improving the user's experience of using the vehicle.

[0007] In some possible implementations, the first usage data includes historical intersection data, historical numerical data, historical classification data and functional data, the historical intersection data is the frequency data of each target function being used when the vehicle is at different historical positions and historical times, the historical numerical data is the frequency data of each target function being used, the historical classification data is the data after classifying the historical positions and the historical times based on preset classification rules, and the functional data is the name information of each target function; obtaining the second usage data based on the first usage data includes: obtaining the second usage data based on the historical intersection data; obtaining the third usage data based on the first usage data includes: obtaining the third usage data based on the historical numerical data, the historical classification data and the functional data.

[0008] In some possible implementations, the method uses a prediction model to predict the target function displayed in the vehicle, and the prediction model includes a first model and a second model; the method also includes: obtaining the current time of the vehicle; obtaining the second usage data based on the historical cross-data includes: inputting the historical cross-data and the current time into the first model to obtain the second usage data; obtaining the third usage data based on the historical numerical data, the historical classification data and the functional data includes: inputting the historical numerical data, the historical classification data, the functional data and the current time into the second model to obtain the third usage data.

[0009] In some possible implementations, the method uses a prediction model to predict the target function displayed in the vehicle, and the prediction model includes a first model and a second model; the method also includes: obtaining the current position of the vehicle; obtaining the second usage data based on the historical intersection data includes: inputting the historical intersection data and the current position into the first model to obtain the second usage data; obtaining the third usage data based on the historical numerical data, the historical classification data and the functional data includes: inputting the historical numerical data, the historical classification data, the functional data and the current position into the second model to obtain the third usage data.

[0010] In some possible implementations, the method uses a prediction model to predict the target function displayed in the vehicle, and the prediction model includes a first model and a second model; the method also includes: obtaining the current time and the current position of the vehicle; the obtaining the second usage data based on the historical intersection data includes: inputting the historical intersection data, the current time and the current position into the first model to obtain the second usage data; the obtaining the third usage data based on the historical numerical data, the historical classification data and the functional data includes: inputting the historical numerical data, the historical classification data, the functional data, the current time and the current position into the second model to obtain the third usage data.

[0011] In some possible implementations, the vehicle state includes a first state, a second state, and a third state; the first state is the state of the vehicle when responding to an unlocking command, the second state is the state of the vehicle during driving, and the third state is the state of the vehicle when parked.

[0012] In a second aspect, the present application discloses a vehicle function recommendation method, which is applied to a vehicle, wherein the vehicle includes multiple target functions, and the method includes: sending the current vehicle state of the vehicle to a cloud server, wherein the cloud server is used to receive first usage data corresponding to each of the target functions, wherein the first usage data is the number of times each of the target functions is used when the vehicle is in the vehicle state during a historical period; obtaining second usage data based on the first usage data, wherein the second usage data represents data on user preference for use of each of the target functions during the historical period; obtaining third usage data based on the first usage data, wherein the third usage data represents predicted data on user preference for use of each of the target functions; calculating usage probability values ​​of each of the target functions based on the second usage data and the third usage data; and determining a target probability value greater than a preset probability threshold among the multiple usage probability values; receiving the target function that matches the target probability value; and displaying a recommendation list, wherein the recommendation list is at least one of the target functions that matches the target probability value determined based on the vehicle state, wherein the recommendation list is the target functions that match the target probability value.

[0013] In some possible implementations, after displaying the recommendation list, the method further includes: in response to a function selection instruction for at least one of the target functions in the recommendation list, executing a function operation corresponding to the selected target function.

[0014] The third aspect of the present application discloses an electronic device, which includes a processor and a memory, wherein the memory is used to store instructions, and the processor is used to call the instructions in the memory so that the electronic device executes the above-mentioned vehicle function recommendation method.

[0015] The fourth aspect of the present application discloses a computer storage medium, including computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes the above-mentioned vehicle function recommendation method.

[0016] It can be understood that the vehicle function recommendation method of the second aspect, the electronic device of the third aspect, and the computer storage medium of the fourth aspect provided above all correspond to the method of the first aspect mentioned above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of a vehicle function recommendation method according to an embodiment of the present application.

[0018] Figure 2This is a schematic diagram of interaction between a vehicle and the cloud according to an embodiment of the present application.

[0019] Figure 3 It is a structural diagram of the Wide&Deep model according to an embodiment of the present application.

[0020] Figure 4 It is a schematic diagram of the architecture of a function recommendation system according to an embodiment of the present application.

[0021] Figure 5 This is another schematic diagram of interaction between a vehicle and the cloud according to an embodiment of the present application.

[0022] Figure 6 It is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application is described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the implementation methods of the present application and the features in the implementation methods can be combined with each other without conflict.

[0024] In the following description, many specific details are set forth to facilitate a full understanding of the present application. The described implementations are only part of the implementations of the present application, rather than all the implementations.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0026] It should be further noted that, in this article, the terms "comprises", "includes" 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 includes 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 presence of other identical elements in the process, method, article or device including the element.

[0027] In this application, "at least one" means one or more, and "more" means two or more than two. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and drawings of this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0028] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0029] Vehicles are equipped with a variety of target functions for different application scenarios. For example, users can massage seats in the vehicle, turn on ambient lights and listen to music while parking, turn on the air conditioner before getting in the car, plan navigation routes in advance, etc. However, it is cumbersome for users to turn on some target functions, resulting in a poor user experience.

[0030] Based on this, please refer to Figure 1 , an embodiment of the present application provides a vehicle function recommendation method, the vehicle includes multiple target functions, the method can combine the vehicle's target functions that are preferred by users in the historical period, and based on the user's historical usage habits of the target functions, recommend target functions that may be of interest to the user, so that the user can directly operate the recommended target function, which increases the convenience of the user's operation of the target function. The vehicle function recommendation method of this embodiment can be applied to a function recommendation system, and the function recommendation system can be deployed in the cloud. The following will be explained by taking the function recommendation system that can be deployed in the cloud as an example. Specifically, the function recommendation system is deployed on a cloud server. The cloud is communicated with the on-board device set in the vehicle. In this embodiment, the on-board device is an on-board computer. In other embodiments, the on-board device can also be other electronic devices besides the on-board computer, as long as it can be communicated with the cloud. The function recommendation system can also be arranged in other devices besides the cloud, and this application does not limit this.

[0031] The cloud is a software platform that uses application virtualization technology, integrating multiple functions such as software search, download, use, management, and backup. Through this platform, all kinds of commonly used software can be encapsulated in an independent virtualized environment, so that the application software will not be coupled with the system, achieving the purpose of green software use. At present, more and more car manufacturers are also choosing to deploy some services in the cloud to achieve seamless connection between multiple devices and between devices and the cloud, shielding the differentiation of terminal capabilities of different devices. Thereby improving the user experience.

[0032] The cloud may include an independent server, or a server network or server cluster composed of servers. For example, the cloud described in the embodiments of the present application includes but is not limited to a computer, a network host, a single network server, a plurality of network server sets or a cloud server composed of multiple servers. The cloud server is composed of a large number of computers or network servers based on cloud computing.

[0033] The vehicle function recommendation method includes the following steps: Step 101: Obtain the current vehicle status of the vehicle.

[0034] In some embodiments, in response to a service request from a vehicle, the service request includes a vehicle status and a vehicle ID. The vehicle ID is used to uniquely identify the vehicle so as to subsequently recommend a target function to the vehicle that sent the service request. Among them, the vehicle status includes a first state, a second state, and a third state; the first state is the state of the vehicle when it responds to an unlocking command, that is, when the vehicle is in the first state, the user can get on and off the vehicle at any time. The second state is the state of the vehicle during driving, and the third state is the state of the vehicle when it is parked, that is, the state of the vehicle when the gear position of the vehicle is in the P gear. In other embodiments, the vehicle status may also include only one state, for example, the vehicle status only includes the first state, and the present application does not limit the type of vehicle status.

[0035] In some embodiments, the current time, current location, environmental data and operating data of the vehicle in different vehicle states can also be obtained. The environmental data includes the temperature inside the vehicle and the temperature outside the vehicle, and the operating data includes the window status, door status, etc. The specific content of the operating data is set according to the actual design requirements.

[0036] For example, when the vehicle is in the first state or the third state, the current time, the current position of the vehicle, environmental data and operating data are obtained, and the current time can be a time point. When the vehicle is in the second state, the current time, the current position of the vehicle, environmental data and operating data are obtained, and the current time can be a time period, and the current position can include multiple positions that match the vehicle's operating trajectory.

[0037] Step 102: Receive first usage data corresponding to each target function.

[0038] In some embodiments, the first usage data is the number of times each target function is used when the vehicle is in different vehicle states during the historical period. The target functions may include music function, navigation function, ambient light mode function, ambient light on function, ambient light off function, seat massage on function, seat massage off function, suspension height setting function, one-button screen off function, rearview mirror folding function, child lock on function, child lock off function, window full close function, window full open function, trunk open function, etc. In this embodiment, the first usage data includes the number of times the user operates one or more target functions when the vehicle is in any of the first state, the second state, or the third state during the historical period, and is located at different historical locations and at different historical times.

[0039] The cloud server stores the first usage data of various vehicles. For example, the first usage data can be stored in a data warehouse. In order to accurately recommend the target functions preferred by the user based on the first usage data, it is necessary to obtain the frequency of use of each target function of the vehicle in the historical period based on the vehicle ID and the current vehicle state of the vehicle. Specifically, when it is detected that the vehicle state is in the first state, the first usage data when the vehicle is in the first state during the historical period is obtained. Similarly, when it is detected that the vehicle state is in the second state, the first usage data when the vehicle is in the second state during the historical period is obtained. When it is detected that the vehicle state is in the third state, the first usage data when the vehicle is in the third state during the historical period is obtained.

[0040] In this embodiment, please refer to Figure 2 The cloud server can use the Kafka message channel to store the first usage data after data preprocessing in the data warehouse. The Kafka message channel has the characteristics of high throughput, distribution, storage persistence, and high availability. At the same time, the first usage data is stored in the cloud data warehouse. In order to continuously use the first usage data to obtain recommended target functions in the future.

[0041] Furthermore, the cloud first performs data preprocessing operations on the first usage data and constructs a prediction model. In this embodiment, the prediction model may be a Wide & Deep model. Data preprocessing includes data analysis, data filtering, data association, and data statistics. Data analysis, data filtering, data association, and data statistics are all prior arts and will not be described in detail here. The specific structure of the Wide & Deep model is described in detail below and will not be described in detail here.

[0042] Step 103: Obtain second usage data according to the first usage data.

[0043] In some embodiments, the second usage data is used to characterize the data of the user's preference for each target function in the historical period. The prediction model includes a first model and a second model, wherein the first model is a Wide model in the Wide&Deep model, and the second model is a Deep model in the Wide&Deep model.

[0044] In order to obtain accurate function recommendation results later, the first usage data and the current time are input into the Wide model to obtain the second usage data. Alternatively, the first usage data and the current position may be input into the Wide model to obtain the second usage data. Alternatively, the first usage data, the current time, and the current position may be input into the Wide model to obtain the second usage data. In other embodiments, the first usage data, the current time, the current position, the environmental data, and the operating data may be input into the Wide model to obtain the second usage data. That is, the first usage data, and one or more of the current time, the current position, the environmental data, and the operating data may be input into the Wide model to obtain the second usage data.

[0045] In this embodiment, please also refer to Figure 3 In order to obtain the second usage data based on the first usage data and one or more of the current time, current location, environmental data and operation data, it is necessary to build a Wide&Deep model first. At the same time, the constructed Wide&Deep model can be stored in the model warehouse. When the Wide&Deep model is needed, the model can be directly called from the model warehouse. Among them, the model warehouse is essentially a database, which is mainly used to store a variety of different constructed models. When using a certain model, it can be directly called from the model warehouse to improve the efficiency of model calling.

[0046] The Wide&Deep model is a recommendation system framework that combines the Wide model and the Deep model, aiming to balance the accuracy and diversity of recommendation results. The model jointly trains the Wide model and the Deep model, using the memory ability of the Wide model (directly learning the "co-occurrence frequency" in historical data) and the generalization ability of the Deep model (the ability to predict new samples and unseen feature combinations), so as to achieve a balance between the accuracy and diversity of the recommendation results.

[0047] Among them, the Wide & Deep model includes the Wide model and the Deep model. The Wide model has memory capabilities and can efficiently implement the memory function by utilizing cross features, thereby making accurate recommendations. It achieves a certain degree of generalization by adding some broad class features, but is limited by the training data and cannot generalize to situations that have never appeared in the training data. The Wide model uses a sparse matrix constructed by the one-hot algorithm to represent features. These features can be single or cross and need to be manually designed.

[0048] The Deep model has the ability to generalize and recommends unseen content by learning low-dimensional dense vectors. However, when the model faces sparse data, it may over-generalize and recommend a lot of irrelevant content, thus affecting accuracy. The Deep model maps the sparse features of the Wide model to the embedding layer and converts them into a 1-dimensional dense matrix to achieve feature extraction and dimensionality reduction.

[0049] In this embodiment, for example, the Wide model hopes to remember which target function has the highest matching degree with the current time and the current location. The Wide model predicts a probability value, that is, the probability that the user will eventually select the recommended target function for the current time and the current location and the recommended target function. If the user eventually selects the recommended target function, then the model will remember that the probability value of the current time and the current location and the recommended target function is very high.

[0050] That is, the first usage data and one or more of the current time, current location, environmental data and operation data are processed by the Wide model to obtain the second usage data, which can improve the memory capacity of the Wide & Deep model.

[0051] Step 104: Obtain third usage data according to the first usage data.

[0052] In some embodiments, the third usage data represents the data that each target function is predicted to be preferred by the user. Similarly, in order to obtain accurate function recommendation results later, the first usage data and the current time are input into the Deep model to obtain the third usage data. Alternatively, the first usage data and the current position may be input into the Deep model to obtain the third usage data. Alternatively, the first usage data, the current time and the current position may be input into the Deep model to obtain the third usage data. In other embodiments, the first usage data, the current time, the current position, the environmental data and the operating data may also be input into the Deep model to obtain the third usage data. That is to say, the first usage data, and one or more of the current time, the current position, the environmental data and the operating data may be input into the Deep model to obtain the third usage data.

[0053] In this embodiment, for example, the Deep model first converts the current time, the current location, and the first usage data into low-dimensional vectors, and searches for data close to the current time and the current location from the first usage data in the vector space, thereby obtaining the third usage data.

[0054] That is, the third usage data is obtained by processing the current time, the current location and the first usage data using the Deep model, which can improve the generalization ability of the Wide & Deep model, so that the Wide & Deep model can recommend unseen content.

[0055] Furthermore, the first usage data includes historical numerical data, historical classification data, functional data and historical cross data. Among them, the historical cross data is the frequency data of each target function being used when the vehicle is at different historical positions and historical times. The historical numerical data is the frequency data of each target function being used. The historical classification data is the data after the historical position and historical time are classified based on the preset classification rules, and the functional data is the name information of each target function.

[0056] In this embodiment, the historical numerical data may include frequency data of the user manipulating one or more target functions at historical times or frequency data of the user manipulating one or more target functions at historical locations.

[0057] In order to obtain historical classification data, it is necessary to set a preset classification rule first. The preset classification rule includes a time classification rule and a location classification rule. Among them, the time classification rule is used to classify the historical time when the vehicle is in the vehicle state. The location classification rule is used to classify the historical location when the vehicle is in the vehicle state. For example, 24 hours a day is divided into six stages: early morning, morning, noon, afternoon, evening and evening, and early morning, morning, noon, afternoon, evening and evening can mutually exclude and comprehensively cover 24 hours a day. Then determine which stage of early morning, morning, noon, afternoon, evening and evening the historical time when the vehicle is in the vehicle state belongs to. Divide the location into six categories: shopping malls, schools, hospitals, banks, bookstores and hotels, and determine which type of shopping malls, schools, hospitals, banks, bookstores and hotels the historical location when the vehicle is in the vehicle state belongs to. In other embodiments, 24 hours a day can be divided into 4 stages, 8 stages, etc. according to actual needs, or the location can be divided into other types, and this application is not limited to this.

[0058] The function data may include list information of target functions when the vehicle is in the first state, list information of target functions when the vehicle is in the second state, and list information of target functions when the vehicle is in the third state. When the vehicle is in the first state, the target functions may include music function, navigation function, ambient light mode function, ambient light on function, ambient light off function, seat massage on function, seat massage off function, suspension height setting function, one-button screen off function, rearview mirror folding function, reading light on function, reading light off function, window opening function, window closing function.

[0059] When the vehicle is in the second state, the target functions may include music function, navigation function, ambient light mode function, ambient light on function, ambient light off function, seat massage on function, seat massage off function, suspension height setting function, one-button screen off function, rearview mirror folding function, reading light on function, reading light off function, window open function, window close function.

[0060] When the vehicle is in the third state, the target functions may include music function, navigation function, ambient light mode function, ambient light on function, ambient light off function, seat massage on function, seat massage off function, suspension height setting function, one-button screen off function, rearview mirror folding function, child lock on function, child lock off function, window full close function, window full open function, trunk open function. In other embodiments, the function data may also include other target functions, for example, the switch of rest mode, the intensity of energy recovery and the selection of driving mode. This application does not limit the scope of function data.

[0061] In this embodiment, historical cross data can be obtained based on historical numerical data, historical classification data and functional data. Historical cross data is a combination of features in different aspects. Take historical cross data as a combination of features of time and location as an example for illustration. Historical cross data may include frequency data of using one or more target functions in the morning when the vehicle is in a shopping mall, frequency data of using one or more target functions in the morning when the vehicle is in a school, frequency data of using one or more target functions at noon when the vehicle is in a shopping mall, etc. In other words, after the data in historical numerical data, historical classification data and functional data are cross-linked with each other, the data obtained is used as historical cross data. In this embodiment, the data in historical numerical data, historical classification data and functional data can be cross-linked with each other artificially, and the data obtained is historical cross data. As long as the frequency information of one or more target functions being used in the historical period at different historical times and historical locations can be reflected from the historical cross data, it is sufficient.

[0062] In some embodiments, the historical intersection data and the current time are input into the Wide model to obtain the second usage data. Alternatively, the historical intersection data and the current position are input into the Wide model to obtain the second usage data. Alternatively, the historical intersection data, the current time and the current position are input into the Wide model to obtain the second usage data. Alternatively, the historical intersection data, the current time, the current position, the environmental data and the operating data are input into the Wide model to obtain the second usage data. That is, the historical intersection data and one or more of the current time, the current position, the environmental data and the operating data are input into the Wide model to obtain the second usage data.

[0063] Please continue reading Figure 3 , the Wide&Deep model includes a feature input layer, an embedding layer, a hidden layer, and an output layer. The historical cross data and one or more of the current time, current position, environmental data, and operation data are input into the feature input layer. After the Wide model processes the data input into the feature input layer, it obtains the second usage data. The second usage data is then directly input into the output layer. In this way, the Wide model can memorize the cross information between the historical position and historical time in the historical cross data and whether each target function is used. For example, when the historical cross data includes that in the morning and when the vehicle is located in a shopping mall (assuming that the vehicle is in the third state at this time), the frequency data of seat massage is 5 times. Then the Wide model can memorize that in the morning and when the vehicle is located in a shopping mall (the vehicle is in the third state), the user prefers to use the target function of seat massage.

[0064] In this embodiment, the Wide model uses the following formula to process one or more of the historical cross data current time, current location, environmental data, and operation data to obtain the second usage data. The formula is as follows: Among them: x represents the data input from the feature input layer to the Wide model, y represents the data of the output layer, θ represents the feature coefficient, and β represents the constant term.

[0065] In some embodiments, historical numerical data, historical classification data, functional data and current time are input into the Deep model to obtain third usage data. Alternatively, historical numerical data, historical classification data, functional data and current position are input into the Deep model to obtain third usage data. Alternatively, historical numerical data, historical classification data, functional data, current time and current position are input into the Deep model to obtain third usage data. Alternatively, historical numerical data, historical classification data, functional data, current time, current position, environmental data and operating data are input into the Deep model to obtain third usage data. Similarly, that is to say, historical numerical data, historical classification data, functional data, and one or more of the current time, current position, environmental data and operating data can be input into the Deep model to obtain third usage data.

[0066] In some embodiments, the historical numerical data, the historical classification data, the functional data, and one or more of the current time, the current location, the environmental data, and the operating data are input into the Deep model to obtain the third usage data, including: vectorizing the historical classification data and one or more of the current time, the current location, the environmental data, and the operating data to obtain a behavior classification vector. Vectorizing the functional data to obtain a function vector. The third usage data is obtained based on the historical numerical data, the behavior classification vector, and the function vector.

[0067] In this embodiment, historical numerical data, historical classification data, functional data, and one or more of the current time, current location, environmental data, and operating data are input into the feature input layer. Since the historical classification data and functional data are discrete data, and the historical numerical data are continuous data, in order to enable the subsequent Deep model to convert the continuous data into a low-dimensional vector, and then based on the processing of the low-dimensional vector, thereby improving the generalization ability of the model, the embedding layer converts the historical classification data, functional data, and one or more of the current time, current location, environmental data, and operating data transmitted by the feature input layer into continuous data.

[0068] Furthermore, the one-hot algorithm is used to convert the historical classification data into continuous data. Among them, the one-hot algorithm is also called one-bit effective encoding, which mainly uses an N-bit state register to encode N states. Each state consists of an independent register bit, and only one bit is valid at any time. The one-hot algorithm is a representation of the classification variable as a binary vector. It is required to map the classification value to an integer value. Then, each integer value is represented as a binary vector. For example: If the time when the vehicle is parked in the historical classification data is "night", this value is not a number and cannot be directly input into the hidden layer, so the historical classification data is processed by one-hot. Based on the time classification rules described above, the value of night is set to 1, and the values ​​of the other five features are 0. In this way, the conversion from non-digital to digital is completed and can be input into the hidden layer. When the running data also needs to be converted into a data type, the data type conversion process of the historical classification data can be referred to. To avoid repetition, it will not be repeated here.

[0069] Since there are many names of target functions in the function data, and there are connections between the target functions. For example, there is a connection between the two target functions "turn on the atmosphere light" and "turn off the atmosphere light". Therefore, the data type conversion rule of the one-hot algorithm is not applicable. The Word2Vec model needs to be used to convert the data type of the function data to reduce the number of features after vectorization and describe the connection between functions. Specifically, the name of each target function is first segmented, for example, "turn on the atmosphere light" is divided into "turn on" and "atmosphere light". After the segmentation, a preset word library is established, that is, the segmented words are grouped together, and then the Word2Vec model is combined with the preset word library to output the word vector corresponding to each function, and the output dimension can be set for the model. For example, "turn on the atmosphere light", the output dimension is 5, and the word vector output of the Word2Vec model for "turn on the atmosphere light" is ['0.12', '0.94', '0.58', '0.39', '0.22']. In this way, the data type conversion is completed, and the converted data is input into the hidden layer.

[0070] In some embodiments, since the historical numerical data is continuous data, the historical numerical data can be directly transmitted to the hidden layer without the need for processing by the embedding layer. The hidden layer includes multiple layers of first activation functions, and the first activation function of each layer is a ReLu activation function. After the hidden layer receives the historical numerical data, the behavior classification vector and the function vector transmitted by the embedding layer, the historical numerical data, the behavior classification vector and the function vector are processed by multiple layers of the first activation function, and the hidden layer can deeply learn the complex relationship between the data to obtain the third usage data.

[0071] Step 105: Calculate the usage probability value of each target function according to the second usage data and the third usage data.

[0072] Specifically, the output layer is provided with a second activation function, and the second activation function is a Sigmoid function. The second usage data and the third usage data are input into the Sigmoid function to calculate a plurality of usage probability values ​​of each target function.

[0073] The output of the Wide & Deep model combines the output of the Wide model and the Deep model, and uses the following formula to obtain multiple usage probability values. The formula is as follows: in, is the Sigmoid function, Represents historical cross data, are the parameters of the Wide model, is the parameter corresponding to the first activation function of the last layer in the Deep model, is the bias parameter.

[0074] In some embodiments, during the training process of the Deep model, the model parameters of the Deep model are optimized using regularization and batch normalization mechanisms, thereby reducing the impact of model overfitting. The formulas for the regularization and batch normalization mechanisms are as follows: in, , , x represents the data input to the Deep model, and y represents the data output by the Deep model. Indicates that the Deep model uses a neural network function, θ represents the parameters in the function, and λ is a hyperparameter that needs to be adjusted according to the model effect. The smaller this hyperparameter is, the simpler the neural network is and the less likely it is to overfit. m is the batch size.

[0075] Step 106: Determine a target probability value that is greater than a preset probability threshold among the multiple usage probability values.

[0076] Among them, the preset probability threshold can be set to 90%, 92% or 95%, which can be set according to actual needs. This application does not limit the specific value of the preset probability threshold.

[0077] Step 107: recommend a target function that matches the target probability value.

[0078] In this embodiment, a target function matching the target probability value is recommended to the vehicle. After receiving the target function matching the target probability value, the vehicle can make a reasonable judgment on the recommended target function in combination with its own vehicle state, thereby obtaining a recommendation list, and recommending the predicted user-preferred recommendation list to the user. The vehicle state also includes the working states of multiple components of the vehicle. The multiple components include a trunk, windows, rearview mirrors, seats, etc.

[0079] For example, the vehicle can make a reasonable judgment on the recommended target function based on its own vehicle status: when the vehicle is in a parking space and there is an obstacle behind the vehicle that prevents the trunk from being opened, the target function of opening the trunk should not be recommended. Or, when the vehicle is parked, many items are piled in the trunk of the vehicle, making it impossible to activate the target function of seat massage for the rear seats. At this time, the target function of activating seat massage for the rear seats should not be recommended.

[0080] In other embodiments, based on the audio and video data played by the user in the historical period, the Wide & Deep model can be used to process the data to recommend audio and video that the user may be interested in. Alternatively, based on the navigation data of the user in the historical period, the Wide & Deep model can be used to process the data to recommend navigation routes to the user.

[0081] Please combine Figure 4 , Figure 4 It includes three aspects: interactive data, offline training and online push. Among them, the interactive data includes the first usage data, the specific vehicle control pool and the prediction model. The specific vehicle control pool is a data pool that matches the vehicle and the vehicle state of the vehicle. The specific vehicle control pool includes the first vehicle control pool, the second vehicle control pool and the third vehicle control pool. The first vehicle control pool stores the first usage data when the vehicle is in the first state, the second vehicle control pool stores the first usage data when the vehicle is in the second state, and the third vehicle control pool stores the first usage data when the vehicle is in the third state. The prediction model includes the constructed Wide&Deep model.

[0082] Offline training includes a big data platform and a training platform. Among them, the big data platform includes a point-of-use data function, a data ETL function, a historical common target function, and a feature data mining function. The point-of-use data function is mainly used to collect the first usage data of the vehicle. The data ETL function is mainly used to preprocess the first usage data. The historical common target function is mainly used to store information on target functions used in the historical period. The feature data mining function is mainly used to mine feature data in the first usage data. The prediction model is also provided in the training platform. The training platform is mainly used to train the prediction model based on the data in the big data platform to improve the recommendation ability of the model.

[0083] The main functions of online push include: when the vehicle is in different vehicle states, the vehicle sends the current time, current location, environmental data and operating data to the cloud server. Based on the vehicle state of the vehicle, the cloud server receives the first usage data when the vehicle is in the vehicle state during the historical period. Then, the first usage data is input into the Wide&Deep model, which calculates multiple usage probability values. Then, the multiple usage probability values ​​are screened. For example, the multiple usage probability values ​​can be sorted and the usage probability values ​​that meet the requirements are used as the target probability values. Finally, the recommended target function is determined according to the target probability value.

[0084] Compared with the related art, the embodiments of the present application have at least the following advantages: By obtaining the current vehicle state of the vehicle, and then receiving the first usage data when the vehicle is in the vehicle state in the historical period, the frequency information of each target function of the vehicle in the historical period is obtained. On the one hand, the first usage data and one or more of the current time, current location, environmental data and operation data are input into the Wide model, so that the Wide model can remember the data of each target function preferred by the user in the historical period, so as to improve the memory ability of the Wide model for the interactive relationship between the data; on the other hand, the first usage data and one or more of the current time, current location, environmental data and operation data are input into the Deep model, so that the Deep model can predict the data of each target function preferred by the user, so as to improve the generalization ability of the Deep model for the interactive relationship between the data. Finally, the usage probability value of each target function is calculated based on the second usage data and the third usage data. Based on the partial usage probability value greater than the preset probability threshold, the target function that meets the user's usage preference is accurately recommended to the user, so that the user can directly use the preferred target function without performing cumbersome operation steps, thereby improving the user's experience of using the vehicle.

[0085] Please refer to Figure 5 , is a schematic diagram of the interaction between the cloud and the vehicle provided in the embodiment of the present application. The specific steps are as follows: Step S1: When the vehicle is in different vehicle states, the vehicle sends a service request to the cloud server.

[0086] In some embodiments, the service request includes a vehicle state and a vehicle ID. The vehicle ID is used to uniquely identify the vehicle. The vehicle state includes a first state, a second state, and a third state. The first state is the state of the vehicle when it responds to the unlocking instruction, that is, when the vehicle is in the first state, the user can get on and off the vehicle at any time. The second state is the state of the vehicle during driving, and the third state is the state of the vehicle when it is parked, that is, the state of the vehicle when the gear position of the vehicle is in the P gear.

[0087] In some embodiments, the vehicle may send one or more of the current time, current location, environmental data, and operating data of the vehicle to the cloud server.

[0088] In this embodiment, the environmental data includes the temperature inside the vehicle and the temperature outside the vehicle, and the operating data may include the status of vehicle windows and vehicle doors, etc. The content of the operating data is set according to actual design requirements.

[0089] Step S2: The cloud server receives first usage data corresponding to each target function, where the first usage data is the number of times each target function is used when the vehicle is in a vehicle state during a historical period.

[0090] Specifically, the first usage data of the vehicle in the vehicle state in the historical period is searched from the data warehouse. The other specific contents of this step are the same as those of step 102, and will not be described here to avoid repetition.

[0091] Step S3: The cloud server uses the constructed Word&Deep model to obtain the second usage data and the third usage data according to the first usage data.

[0092] In some embodiments, the cloud server can also use the constructed Word&Deep model to obtain the second usage data and the third usage data according to the first usage data and one or more of the current time, current location, environmental data and operation data. The cloud server calls the constructed Word&Deep model from the model warehouse. The specific content of the cloud server using the Word&Deep model to process data is the same as step 103 and step 104, and will not be repeated here to avoid repetition.

[0093] Step S4: The cloud server calculates the usage probability value of each target function according to the second usage data and the third usage data.

[0094] The specific content of this step is the same as step 105, and will not be repeated here to avoid repetition.

[0095] Step S5: The cloud server determines a target probability value that is greater than a preset probability threshold among the multiple usage probability values.

[0096] The specific content of this step is the same as step 106, and will not be repeated here to avoid repetition.

[0097] Step S6: The cloud server sends the target function matching the target probability value to the vehicle.

[0098] Step S7: The vehicle determines a recommendation list according to the vehicle status and displays the recommendation list.

[0099] In this embodiment, after the vehicle receives the target function that matches the target probability value sent by the cloud server, the vehicle can determine a recommendation list according to the vehicle state. The recommendation list is a list of information of multiple target functions. And based on the recommendation list, it recommends to the user. The vehicle state also includes the working state of multiple components of the vehicle. The multiple components include the trunk, windows, rearview mirrors, seats, etc.

[0100] For example, the vehicle can make a reasonable judgment on the target function that matches the target probability value based on its own vehicle state. This means that when the vehicle is in a parking space and there is an obstacle behind the vehicle that prevents the trunk from being opened, the target function of opening the trunk should not be recommended. Or, when the vehicle is parked, many items are piled in the trunk of the vehicle, making it impossible to activate the target function of seat massage for the rear seats. In this case, the target function of activating seat massage for the rear seats should not be recommended.

[0101] Compared with the related art, the embodiments of the present application have at least the following advantages: By obtaining the current vehicle state of the vehicle, and then receiving the first usage data when the vehicle is in the vehicle state in the historical period, the frequency information of each target function of the vehicle in the historical period is obtained. On the one hand, the first usage data and one or more of the current time, current location, environmental data and operation data are input into the Wide model, so that the Wide model can remember the data of each target function preferred by the user in the historical period, so as to improve the memory ability of the Wide model for the interactive relationship between the data; on the other hand, the first usage data and one or more of the current time, current location, environmental data and operation data are input into the Deep model, so that the Deep model can predict the data of each target function preferred by the user, so as to improve the generalization ability of the Deep model for the interactive relationship between the data. Finally, the usage probability value of each target function is calculated based on the second usage data and the third usage data. Based on the partial usage probability value greater than the preset probability threshold, the target function that meets the user's usage preference is accurately recommended to the user, so that the user can directly use the preferred target function without performing cumbersome operation steps, thereby improving the user's experience of using the vehicle.

[0102] The embodiment of the present application also provides a vehicle function recommendation method, which is applied to a vehicle, and includes: sending the current vehicle state of the vehicle to a cloud server. The cloud server is used to receive first usage data corresponding to each target function, and the first usage data is the number of times each target function is used when the vehicle is in the vehicle state in a historical period.

[0103] Furthermore, the cloud server is also used to obtain second usage data and third usage data based on the first usage data, the second usage data represents the data of each target function preferred by users in a historical period, and the third usage data represents the predicted data of each target function preferred by users.

[0104] The cloud server is also used to calculate the usage probability value of each target function according to the second usage data and the third usage data, and to determine a target probability value greater than a preset probability threshold among the multiple usage probability values.

[0105] The vehicle is used to receive a target function that matches a target probability value and display a recommendation list, wherein the recommendation list is at least one target function that matches the target probability value determined based on the vehicle state. In this embodiment, after the vehicle receives the target function that matches the target probability value sent by the cloud server, the vehicle can determine the recommendation list according to the vehicle state.

[0106] Furthermore, after the vehicle displays the recommendation list, the vehicle responds to the function selection instruction for the recommendation list and executes the function operation corresponding to the selected target function. For example, the recommendation list includes turning on the reading light function, turning off the ambient light function and closing the window function, and the user can select one or more of the turning on the reading light function, turning off the ambient light function and closing the window function. If the user selects turning on the reading light function and turning off the ambient light function, the vehicle controls the reading light to be turned on and the ambient light to be turned off.

[0107] Compared with the related art, the embodiments of the present application have at least the following advantages: By obtaining the current vehicle state of the vehicle and then receiving the first usage data when the vehicle was in the vehicle state in the historical period, the frequency information of the use of each target function of the vehicle in the historical period is obtained. The first usage data is input into the prediction model to obtain the second usage data and the third usage data. Then, based on the second usage data and the third usage data, the usage probability value of each target function is calculated. Based on the partial usage probability values ​​greater than the preset probability threshold, the target function that meets the user's usage preference is accurately recommended to the user, so that the user can directly use the preferred target function without performing cumbersome operation steps, thereby improving the user's experience of using the vehicle.

[0108] Please refer to Figure 6 , is a schematic diagram of the hardware structure of the electronic device 1000 provided in the embodiment of the present application. Figure 6As shown, the electronic device 1000 may include a processor 1001 and a memory 1002. The memory 1002 is used to store one or more computer programs 1003. The one or more computer programs 1003 are configured to be executed by the processor 1001. The one or more computer programs 1003 include instructions, and the above instructions can be used to implement the above method in the electronic device 1000.

[0109] It is understandable that the structure shown in this embodiment does not constitute a specific limitation on the electronic device 1000. In other embodiments, the electronic device 1000 may include more or fewer components than shown, or combine or separate some components, or arrange the components differently.

[0110] The processor 1001 may include one or more processing units, for example, the processor 1001 may include an application processor (AP), a modem, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Different processing units may be independent devices or integrated into one or more processors.

[0111] The processor 1001 may also be provided with a memory for storing instructions and data. In some embodiments, the memory in the processor 1001 is a cache memory. The memory may store instructions or data that the processor 1001 has just used or circulated. If the processor 1001 needs to use the instruction or data again, it may be directly called from the memory. This avoids repeated access, reduces the waiting time of the processor 1001, and thus improves the efficiency of the system.

[0112] In some embodiments, the processor 1001 may include one or more interfaces. The interface may include an inter-integrated circuit (I2C) interface, an inter-integrated circuit sound (I2S) interface, a pulse code modulation (PCM) interface, a universal asynchronous receiver / transmitter (UART) interface, a mobile industry processor interface (MIPI), a general-purpose input / output (GPIO) interface, a SIM interface, and / or a USB interface, etc.

[0113] In some embodiments, processor 1001 is used to execute acceleration schemes such as single instruction multiple data (SIMD) and very long instruction word (VLIW).

[0114] In some embodiments, memory 1002 may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0115] This embodiment also provides a computer-readable storage medium, in which computer instructions are stored. When the instructions are executed on an electronic device, the electronic device executes the above-mentioned related method steps to implement the vehicle function recommendation method in the above-mentioned embodiment.

[0116] Among them, the electronic device and computer-readable storage medium provided in this embodiment are used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0117] In practical applications, the above functions can be distributed to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0118] In several embodiments provided in the present application, the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are schematic. For example, the division of the modules or units is 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 device, 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 through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0119] The unit described as a separate component may or may not be physically separated, and the component shown as a unit may be one physical unit or multiple physical units, that is, it may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment.

[0120] In addition, each functional unit in each embodiment of the present application 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.

[0121] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program code.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the above embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present application.

Claims

1. A vehicle function recommendation method, characterized in that: The vehicle includes a plurality of target functions, and the method includes: Get the current vehicle status of the vehicle; receiving first usage data corresponding to each of the target functions, the first usage data being the number of times each of the target functions is used when the vehicle is in the vehicle state during a historical period; Obtaining second usage data according to the first usage data, wherein the second usage data represents data on the preferred usage of each of the target functions by users during the historical period; Obtaining third usage data according to the first usage data, wherein the third usage data represents data on predicted user preference for each of the target functions; Calculating a usage probability value of each of the target functions according to the second usage data and the third usage data; Determine a target probability value that is greater than a preset probability threshold value among the plurality of usage probability values; The target function matching the target probability value is recommended.

2. The vehicle function recommendation method according to claim 1, characterized in that: The first usage data includes historical cross data, historical numerical data, historical classification data and function data, wherein the historical cross data is the frequency data of each target function being used when the vehicle is at different historical positions and historical times, the historical numerical data is the frequency data of each target function being used, the historical classification data is the data after the historical positions and historical times are classified based on preset classification rules, and the function data is the name information of each target function; The obtaining second usage data according to the first usage data comprises: Obtaining the second usage data according to the historical cross data; The obtaining the third usage data according to the first usage data comprises: The third usage data is obtained according to the historical numerical data, the historical classification data and the functional data.

3. The vehicle function recommendation method according to claim 2, characterized in that: The method uses a prediction model to predict the target function displayed in the vehicle, the prediction model including a first model and a second model; the method also includes: Get the current time of the vehicle; The obtaining the second usage data according to the historical cross data includes: Inputting the historical cross-data and the current time into the first model to obtain the second usage data; The obtaining the third usage data according to the historical numerical data, the historical classification data and the functional data comprises: The historical numerical data, the historical classification data, the functional data and the current time are input into the second model to obtain the third usage data.

4. The vehicle function recommendation method according to claim 2, characterized in that: The method uses a prediction model to predict the target function displayed in the vehicle, the prediction model including a first model and a second model; the method also includes: Obtaining the current position of the vehicle; The obtaining the second usage data according to the historical cross data includes: Inputting the historical intersection data and the current position into the first model to obtain the second usage data; The obtaining the third usage data according to the historical numerical data, the historical classification data and the functional data comprises: The historical numerical data, the historical classification data, the functional data and the current position are input into a second model to obtain the third usage data.

5. The vehicle function recommendation method according to claim 2, characterized in that: The method uses a prediction model to predict the target function displayed in the vehicle, the prediction model including a first model and a second model; the method also includes: Obtain the current time and current position of the vehicle; The obtaining the second usage data according to the historical cross data includes: Inputting the historical intersection data, the current time, and the current location into the first model to obtain the second usage data; The obtaining the third usage data according to the historical numerical data, the historical classification data and the functional data comprises: The historical numerical data, the historical classification data, the functional data, the current time and the current location are input into the second model to obtain the third usage data.

6. The vehicle function recommendation method according to claim 1, characterized in that: The vehicle state includes a first state, a second state and a third state; the first state is the state of the vehicle when responding to an unlocking instruction, the second state is the state of the vehicle during driving, and the third state is the state of the vehicle when it is parked.

7. A vehicle function recommendation method, characterized in that: Applied to a vehicle, the vehicle comprising a plurality of target functions, the method comprising: The current vehicle state of the vehicle is sent to a cloud server, wherein the cloud server is used to receive first usage data corresponding to each of the target functions, the first usage data being the number of times each of the target functions is used when the vehicle is in the vehicle state during a historical period; second usage data is obtained based on the first usage data, the second usage data representing data of user preference for use of each of the target functions during the historical period; third usage data is obtained based on the first usage data, the third usage data representing predicted data of user preference for use of each of the target functions; usage probability values ​​of each of the target functions are calculated based on the second usage data and the third usage data; and a target probability value greater than a preset probability threshold is determined among the multiple usage probability values; receiving the target function matched with the target probability value; A recommendation list is displayed, wherein the recommendation list is at least one of the target functions that matches the target probability value and is determined based on the vehicle state.

8. The vehicle function recommendation method according to claim 7, characterized in that: After displaying the recommendation list, the method further includes: In response to a function selection instruction for at least one of the target functions in the recommendation list, a function operation corresponding to the selected target function is executed.

9. An electronic device, characterized in that: The electronic device comprises a processor and a memory, wherein the memory is used to store instructions, and the processor is used to call the instructions in the memory so that the electronic device executes the vehicle function recommendation method according to any one of claims 1 to 8.

10. A computer storage medium, characterized in that: The method comprises computer instructions, and when the computer instructions are executed on an electronic device, the electronic device executes the vehicle function recommendation method according to any one of claims 1 to 8.

Images