Self-adaptive automobile electronic control system based on deep learning

Abstract

The invention discloses a self-adaptive automobile electronic control system based on deep learning, and relates to the technical field of automobile electronic control, and the self-adaptive automobile electronic control system comprises a data acquisition module, a data analysis and processing module and a user behavior operation habit prediction module. And obtaining an operation behavior habit data set of the user based on the use data, generating a personalized recommendation system of the user based on the operation behavior habit data set of the user, performing real-time monitoring based on the generated personalized recommendation system of the user, and judging the feasibility and the matching degree of the personalized recommendation system of the user. The scheme has the advantages that the operation behavior habit data set of the user is acquired, the personalized recommendation system is generated, the vehicle use experience of the user is improved, the favor degree of the user is increased, a self-adaptive automobile electronic control system is enhanced, the favor of an automobile company is expanded, and customers are attracted to purchase vehicles.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of automobile electronic control, in particular to an adaptive automobile electronic control system based on deep learning. BACKGROUND

[0002] The adaptive automobile electronic control refers to real-time analysis and adjustment of parameters and strategies of an automobile electronic system through deep learning technology, so that the automobile electronic system can intelligently optimize control behavior according to different driving environments, road conditions and driver habits, so as to improve driving safety, comfort and efficiency.

[0003] With the development of electric vehicle technology, the electric control system in the automobile has become an indispensable core technology of the battery, and the electric equipment in the electric vehicle has occupied the main part of the overall electric energy consumption. However, due to the difference in user behavior habits, the electric energy consumption in the electric vehicle cannot always be in the optimal state, resulting in low overall electric energy utilization rate. SUMMARY

[0004] To solve the above technical problems, the adaptive automobile electronic control system based on deep learning is provided, which solves the above problems that with the development of electric vehicle technology, the electric control system in the automobile has become an indispensable core technology of the battery, and the electric equipment in the electric vehicle has occupied the main part of the overall electric energy consumption. However, due to the difference in user behavior habits, the electric energy consumption in the electric vehicle cannot always be in the optimal state, resulting in low overall electric energy utilization rate.

[0005] To achieve the above purposes, the technical scheme adopted by the application is as follows:

[0006] The adaptive automobile electronic control system based on deep learning comprises:

[0007] A data acquisition module acquires historical operation behavior data of automobile users, constructs an operation behavior habit data set of the users, and the operation behavior data comprises:

[0008] A data analysis and processing module analyzes the trend change of the operation behavior habit based on the operation behavior habit data set of the users, and generates personalized demands of the users;

[0009] A user behavior operation habit prediction module acquires spatial environment data in the automobile based on a heterogeneous sensor, analyzes the matching degree between the current spatial environment data and the personalized demands of the users, and generates an automobile electronic control scheme.

[0010] Preferably, the data acquisition module acquires historical operation behavior data of automobile users, constructs an operation behavior habit data set of the users, and the operation behavior data comprises:

[0011] The navigation function uses data, vehicle-mounted air conditioner data, temperature sensor data, automobile steering wheel heating data, automobile seat massage data, automobile interior lamp data, and analyzes the data;

[0012] Based on the data analysis, the different device categories of the automobile are mapped with the operation behavior instructions to obtain a set matrix B of automobile devices;

[0013]

[0014] wherein b ch represents the operation behavior h of the cth device category, i is the total number of device categories, and j is the total number of operation behaviors;

[0015] Based on the set matrix B of automobile devices, the operation behavior habit data set of the user is obtained.

[0016] Preferably, the data analysis and processing module analyzes the trend change of the operation behavior habit based on the operation behavior habit data set of the user, and generates the personalized demand of the user;

[0017] The recent operation behavior data of the user is obtained, and the set matrix B of automobile devices is used for verification to determine whether the operation behavior habit of the user has changed. If not, it is not objectionable, and if so, it is recorded as a user to be changed.

[0018] Based on the user to be changed, the change of the operation behavior habit of the user to be changed is obtained, and the microcontroller is used to record and update the storage memory to update the operation behavior habit data set of the user in real time;

[0019] Based on the implementation of updating the operation behavior habit data set of the user, the personalized demand of the user is generated.

[0020] Preferably, the user behavior operation habit prediction module obtains the spatial environment data inside the automobile based on the heterogeneous sensor, analyzes the matching degree between the current spatial environment data and the personalized demand of the user, and generates an electronic control scheme of the automobile;

[0021] Based on the heterogeneous sensor, the spatial environment data inside the automobile is obtained, and the spatial environment data inside the automobile specifically includes:

[0022] The use frequency data of the devices of the automobile (navigation function use data, vehicle-mounted air conditioner, automobile seat massage, automobile steering wheel heating, and automobile interior lamp), use time data (morning and evening peak, weekend, and holiday), and use scene data (city commuting, long-distance travel, and short-distance travel);

[0023] Based on the automobile device use data of the user, features are extracted, and the extracted features specifically include:

[0024] The average number of times of using each automobile device per week, the proportion of using each device in a certain period of time, and the seasonal change of function use;

[0025] Based on the extracted features, a decision tree model is constructed according to binary classification and selection of information gain as the optimal attribute;

[0026] The decision tree model, the sample set is denoted as D, after being divided by the optimal attribute, the sample set is divided into v subsets, D1, D2,..., Dv, the information entropy of D before and after division and the information gain are calculated. *

[0027] The information entropy formula of D before division is:

[0028]

[0029] In the formula, y represents the result, and Pn represents the proportion of the number of samples of the nth category to the total number of samples D.

[0030] The information entropy formula of the sample set after division is:

[0031]

[0032] In the formula, v represents the number of values that the attribute can take, if v represents the use time, then v=3, D n represents the number of samples of the nth value of the attribute.

[0033] The information gain formula is:

[0034] Gain(D, a) = H(D) - H(D * ) 划分后

[0035] In the formula, a represents the attribute used in the division.

[0036] Based on the information gain, the information gain of each attribute before division is calculated, and the attribute with the maximum information gain is selected as the optimal division attribute.

[0037] Based on the optimal attribute, the use frequency of the navigation function is taken as the starting point, the data set is divided into different subsets according to the attribute in the internal node, the optimal attribute is continuously selected for division, and the leaf node stores the prediction result, i.e., the personalized demand recommendation of the user, when the stop condition is reached.

[0038] Based on the decision tree model, specifically comprising:

[0039] ​With the navigation function use frequency, it is judged whether the user is a high-frequency use, if yes, the use scene of the user high-frequency navigation is judged and the corresponding recommendation is carried out (long-distance travel recommends senior navigation package, city commuting recommends city commuting optimization navigation), the use time is judged after the use scene is judged;

[0040] Based on the judgment of whether the user is a high-frequency use, if not, the frequently used equipment in the user's car is judged according to the obtained microcontroller data, and intelligent recommendation is carried out.

[0041] Compared with the prior art, the beneficial effects of the present application are:

[0042] The present application proposes a self-adaptive automobile electronic control system based on deep learning, obtains user operation behavior habit data set through matrix model, generates user personalized automobile recommendation system, predicts the feasibility of personalized system through decision tree model, improves user's driving experience, brings convenience for user's driving behavior, upgrades the intelligent optimization system of automobile, expands the reputation of the company. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 It is a method flow chart of the self-adaptive automobile electronic control system based on deep learning;

[0044] Figure 2 It is a method flow chart for constructing user operation behavior habit data set;

[0045] Figure 3 It is a method flow chart for generating user's personalized demand;

[0046] Figure 4 It is a method flow chart for generating user's personalized demand. DETAILED DESCRIPTION

[0047] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be thought by those skilled in the art.

[0048] Reference Figure 1 As shown, the self-adaptive automobile electronic control system based on deep learning comprises:

[0049] The data acquisition module acquires the historical operation behavior data of the automobile user, constructs the operation behavior habit data set of the user, and the operation behavior data comprises:

[0050] The data analysis and processing module analyzes the trend change of the operation behavior habit based on the operation behavior habit data set of the user, and generates the personalized demand of the user;

[0051] The user behavior operation habit prediction module, based on a heterogeneous sensor, implements obtaining space environment data in the car interior, analyzes the matching degree between the current space environment data and the personalized needs of the user, and generates a car electronic control scheme.

[0052] The scheme mines the behavior preferences of the user through deep learning of the adaptive electronic control system of the car, formulates the car electronic control scheme of the user, improves the car use experience of the user, and brings convenience to the user's life.

[0053] Referring to Figure 2 The data acquisition module acquires historical operation behavior data of the car user, constructs an operation behavior habit data set of the user, and the operation behavior data includes:

[0054] Navigation function use data, car-mounted air conditioner data, temperature sensor data, car steering wheel heating data, car seat massage data, and car interior lamp data are analyzed and arranged.

[0055] Based on the analysis and arrangement of the data, different device categories of the car are mapped with operation behavior instructions to obtain a device use set matrix B of the car.

[0056]

[0057] Wherein, b ch represents the operation behavior h when the cth device category, i is the total number of device categories, and j is the total number of operation behaviors.

[0058] Based on the device use set matrix B of the car, an operation behavior habit data set of the user is obtained.

[0059] By acquiring the device use data of the car, the operation behavior habit data set of the user is obtained by means of matrix model analysis, and the operation behavior habit data set is sent to the microcontroller in the electronic control unit (ECU) by means of the CAN communication module. The microcontroller stores and remembers.

[0060] Referring to Figure 3 The data analysis and processing module, based on the operation behavior habit data set of the user, analyzes the trend change of the operation behavior habit, and generates the personalized needs of the user.

[0061] The recent operation behavior data of the user is acquired, the device use set matrix B of the car is used for verification, and it is judged whether the operation behavior habit of the user has changed. If not, it is not objectionable, and if not, it is recorded as a user to be changed.

[0062] Based on the record of the user to be changed, the operation behavior habit change of the user to be changed is acquired, and the microcontroller is used to record and update the storage memory, and the operation behavior habit data set of the user is updated in real time.

[0063] Based on the operation behavior habit data set of the user, the personalized demand of the user is generated.

[0064] Referring to Figure 4 The user behavior operation habit prediction module is based on a heterogeneous sensor, acquires spatial environment data inside the car, analyzes the matching degree between the current spatial environment data and the personalized demand of the user, and generates the personalized demand of the user.

[0065] Based on the heterogeneous sensor, the spatial environment data inside the car is acquired, and the spatial environment data inside the car specifically includes:

[0066] The use frequency data of each device of the car (navigation function use data, car-mounted air conditioner, car seat massage, car steering wheel heating, car interior lamp), use time data (morning and evening peak, weekend, holiday), use scene data (city commuting, long-distance travel, short-distance travel);

[0067] Based on the use data of each device of the car of the user, the features are extracted, and the extracted features specifically include:

[0068] The average number of times of using a certain car device per week, the proportion of using each device in a certain time period, and the seasonal change of function use;

[0069] Based on the extracted features, a decision tree model is constructed according to the binary classification and the optimal attribute selection information gain;

[0070] The decision tree model, the sample set is denoted as D, after the optimal attribute division, the sample set is divided into v subsets D1, D2,..., Dv, and the information entropy and the information gain of D before division and D * after division are calculated;

[0071] The information entropy formula of D before division is:

[0072]

[0073] In the formula, y represents the result, and Pn represents the proportion of the number of samples of the nth category to the total number of samples D;

[0074] The information entropy formula of the sample set after division is:

[0075]

[0076] In the formula, v represents the number of values that the attribute can take, if v represents the use time, then v=3, and D n represents the number of samples of the nth value of the attribute;

[0077] The information gain formula is:

[0078] Gain(D, a) = H(D) - H(D * ) 划分后

[0079] In the formula, a represents the attribute used in the division;

[0080] Based on the information gain, the information gain of each attribute is calculated before the division, and the attribute with the maximum information gain is selected as the optimal division attribute;

[0081] Based on the optimal division attribute, the navigation function usage frequency is taken as the starting point, the internal node divides the data set into different subsets according to the attribute, and the optimal attribute is continuously selected for segmentation, and the leaf node stores the prediction result, i.e. the personalized demand recommendation of the user, when the stop condition is reached;

[0082] Based on the decision tree model, specifically comprising:

[0083] The navigation function usage frequency is used to determine whether the user is a high-frequency user, and if so, the use scenario of the user's high-frequency navigation is determined and the corresponding recommendation is made (high-end navigation package is recommended for long-distance travel, and city commuting optimized navigation is recommended for city commuting). After determining the use scenario, the use time is determined;

[0084] Based on the judgment of whether the user is a high-frequency user, if not, the frequently used devices in the user's car are determined according to the obtained microcontroller data, and intelligent recommendation is made.

[0085] Through the decision tree model, the use relationship of each device in the user's car is obtained, and the personalized demand of the user is intelligently recommended according to the relationship between the use of each device in the user's car, the matching degree between the personalized demand of the user is determined, and the automobile electronic control scheme is generated.

[0086] In summary, the adaptive automobile electronic control system based on deep learning can not only understand the preferences of the target car owner in a timely manner, but also can change the instructions immediately when the target car owner's preferences change, improve the target car owner's experience and good feeling, and bring reputation to the car company.

[0087] Compared with the prior art, the adaptive automobile electronic control system based on deep learning has the beneficial effects that by collecting the use data of the user on the car equipment, combining the matrix model to obtain the operation behavior habit data set of the user, and then detecting the feasibility and matching degree of the behavior operation habit data set in real time, a user personalized intelligent recommendation system is generated, which not only saves time for the user and brings convenience to the user's driving behavior, improves the user's experience and good feeling, but also continuously upgrades the automobile electronic control system, keeps pace with the progress of the times, and gains reputation for the car company, attracting customers.

[0088] In summary, the application has the advantages of keeping pace with the times, upgrading the automobile electronic control system, generating personalized recommendation data of users, enhancing the favorability of automobile users, bringing convenience to the use process of automobile users, enhancing the experience of users, creating a good reputation for automobile companies, expanding the popularity, attracting users, and improving the car purchase behavior of users.

[0089] The basic principles, main features and advantages of the application are shown and described above. Those skilled in the art should understand that the application is not limited by the above examples, and the above examples and descriptions in the specification are only the principles of the application. Without departing from the spirit and scope of the application, various changes and improvements can be made to the application, and these changes and improvements all fall within the scope of the claimed application. The scope of protection required by the application is defined by the appended claims and their equivalents.

Claims

1. A deep learning based adaptive automotive electronic control system, characterized by, include: The data acquisition module acquires historical operation behavior data of car users and constructs a user operation behavior habit dataset, which includes: The data analysis and processing module analyzes the trend changes in user operation behavior habits based on the user's operation behavior habit dataset and generates personalized user needs; The user behavior and operation habit prediction module, based on heterogeneous sensors, acquires spatial environment data inside the car, analyzes the matching degree between the current spatial environment data and the user's personalized needs, and generates a car electronic control scheme.

2. The deep learning based adaptive automotive electronic control system of claim 1, wherein, Specifically, this includes: a data acquisition module, which acquires historical operational behavior data of car users and constructs a user operational behavior habit dataset, wherein the operational behavior data includes: Data on navigation function usage, vehicle air conditioning, temperature sensor data, vehicle steering wheel heating, vehicle seat massage, and vehicle interior lighting are collected, organized, and analyzed. Based on the data processing and analysis, different vehicle equipment categories are mapped to operation commands to obtain the vehicle equipment usage set matrix B. wherein b ch h, i, j represent the operating behavior of the cth device category, i is the total number of device categories, and j is the total number of operating behaviors. Based on the set matrix B of the various devices in the car, a dataset of user operating behavior habits is obtained.

3. The deep learning adaptive automotive electronic control system of claim 2, wherein, The data analysis and processing module analyzes the trend changes in user operation behavior habits based on the user's operation behavior habit dataset and generates personalized user needs; Obtain recent user behavior data and use the set matrix B of various car devices to verify whether the user's operating behavior habits have not changed. If so, there is no objection; otherwise, record the user as a user to be changed. Based on the record of users to be changed, data analysis is performed to identify changes in the user's operational behavior habits. The microcontroller is used to record, update, and store the data, and the user's operational behavior habit dataset is updated in real time. Based on the updated user behavior dataset, personalized user needs are generated.

4. The deep learning adaptive automotive electronic control system of claim 3, wherein, The user behavior and operation habit prediction module, based on heterogeneous sensors, acquires spatial environment data inside the car, analyzes the matching degree between the current spatial environment data and the user's personalized needs, and generates a car electronic control scheme. Based on heterogeneous sensors, spatial environmental data inside a vehicle is acquired, specifically including: Data on the frequency of use of various car devices (navigation function usage data, car air conditioning, car seat massage, car steering wheel heating, car interior lights), usage time data (morning and evening rush hours, weekends, holidays), and usage scenario data (city commuting, long-distance travel, short-distance travel). Based on the user's vehicle usage data for various devices, features are extracted, specifically including: Average number of times a car device is used per week, percentage of each device's usage within a specific time period, and seasonal variations in function usage; Based on the extracted features, a decision tree model is constructed according to binary classification and information gain as the optimal attribute.

5. The deep learning adaptive automotive electronic control system of claim 4, wherein, The decision tree model, sample set is denoted as D, after optimal attribute division, the sample set is divided into v subsets, D1, D2,..., Dv, the information entropy and information gain of D before division and D after division are calculated; * ​ The formula for the information entropy of D before partitioning is: In the formula, y represents the result, and Pn represents the proportion of the number of samples in the nth category to the total number of samples D; The formula for the information entropy of the partitioned subsample set is: where v represents the number of values that the attribute can take, and if v represents the time of use, then v = 3, D n represents the number of samples for the nth value of the attribute. The information gain formula is: Gain(D, a) = H(D) - H(D * ) 划分后 In the formula, a represents the attribute used in the present division.

6. A method according to claim 5, wherein, Based on information gain, calculate information gain for each attribute before division, select the attribute with the maximum information gain as the optimal division attribute; Based on the optimal division attribute, take the navigation function use frequency as the starting point, the internal node divides the data set into different subsets according to the attribute, continues to select the optimal attribute for segmentation, and the leaf node stores the prediction result, that is, the personalized demand recommendation of the user, when the stop condition is reached; Based on the decision tree model, specifically comprising: With the navigation function use frequency, judge whether the user is a high-frequency user, if yes, judge the use scene of the user's high-frequency navigation and make corresponding recommendations (recommend senior navigation package for long-distance travel, recommend city commuting optimization navigation for city commuting), and judge the use time after judging the use scene; Based on the judgment of whether the user is a high-frequency user, if not, according to the obtained microcontroller data, judge the frequently used devices in the user's car and make intelligent recommendations.

Images