A driving mode recognition method, device, terminal device and storage medium

Abstract

The present invention discloses a driving mode recognition method, device, terminal equipment and storage medium, the method comprising: obtaining driver data, vehicle data and environmental data during vehicle driving; performing data preprocessing on the driver data, vehicle data and environmental data; extracting and analyzing the preprocessed data by feature engineering technology, and selecting effective feature vectors that affect the driving mode by PCA dimensionality reduction technology; establishing a driving mode recognition model according to the effective feature vectors that affect the driving mode, and training and verifying the driving mode recognition model to obtain a driving mode recognition result. The embodiment of the present invention is based on a large amount of data during vehicle driving, can accurately reflect the actual situation of the vehicle driving process, and effectively improve the accuracy of driving mode recognition.

Description

Technical Field

[0001] The present invention relates to the field of vehicle intelligent interaction technology, and in particular to a driving mode recognition method, device, terminal equipment and storage medium. Background Art

[0002] With the development of electric vehicles’ intelligence, people’s demand for a good car experience makes people hope that cars can understand them more and more, and can customize corresponding service content and assisted driving according to their status and needs. Therefore, current electric vehicles are basically equipped with driving assistance systems, ACC and other automotive electronic systems to improve driving safety and comfort. The driving assistance system generally needs to detect and analyze the vehicle’s driving condition data and the driver’s driving data to identify the driver’s driving mode, so as to provide the driver with more humane services and safer and more comfortable assisted driving.

[0003] However, current methods for identifying the driver's driving mode usually only use a single vehicle condition data to identify the driver's driving mode. This recognition method based on a small amount of data cannot accurately reflect the actual situation of the vehicle driving process, resulting in a low accuracy rate in identifying the driver's driving mode. Summary of the invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a driving mode recognition method, apparatus, terminal device and storage medium, which can effectively improve the accuracy of driving mode recognition by using an improved SOM algorithm based on a large amount of data during vehicle driving.

[0005] In order to achieve the above object, an embodiment of the present invention provides a vehicle driving mode recognition method, comprising:

[0006] Acquiring driver data, vehicle data and environmental data during vehicle driving; wherein the driver data includes driver status data and driver behavior data, the vehicle data includes vehicle operating condition data, battery data and time data, and the environmental data includes climate data and geographic data;

[0007] performing data preprocessing on the driver data, the vehicle data and the environment data;

[0008] Feature engineering technology is used to extract and analyze the preprocessed data, and PCA dimensionality reduction technology is used to select effective feature vectors that affect driving patterns;

[0009] A driving mode recognition model is established according to the effective feature vector that affects the driving mode, and the driving mode recognition model is trained and verified to obtain a driving mode recognition result.

[0010] As an improvement of the above scheme, the driver status data includes fatigue level, age and gender; the driver behavior data includes throttle data, braking data and sharp turn data; the vehicle operating condition data includes average vehicle speed, vehicle speed extremes, tire pressure and tire temperature; the battery data includes current, voltage and battery temperature; the time data includes single driving time and timestamp; the climate data includes temperature, humidity, season and light; the geographic data includes location, road conditions and terrain.

[0011] As an improvement of the above solution, the acquisition of driver data, vehicle data and environmental data during vehicle driving specifically includes:

[0012] Acquire VDCM domain data, BDCM domain data and IDCM domain data of the vehicle through the big data platform; wherein VDCM refers to signals related to vehicle chassis power, BDCM refers to signals related to the vehicle body, and IDCM refers to signals related to the smart cockpit;

[0013] The VDCM domain data, the BDCM domain data and the IDCM domain data are screened to obtain the driver data, the vehicle data and the environment data during vehicle driving.

[0014] As an improvement of the above solution, the data preprocessing of the driver data, the vehicle data and the environment data includes at least one of the following:

[0015] removing noise data and outliers from the driver data, the vehicle data, and the environment data;

[0016] For missing values ​​and invalid values ​​in the driver data, the vehicle data and the environment data, the adjacent average values ​​are selected for filling;

[0017] Performing logarithmic transformation on the relatively scattered data among the driver data, the vehicle data and the environment data;

[0018] Performing binning processing on the data according to data characteristics of the driver data, the vehicle data, and the environment data;

[0019] The driver data, the vehicle data and the environment data are standardized and normalized.

[0020] As an improvement to the above solution, a driving mode recognition model is established, specifically: a driving mode recognition model based on an improved SOM algorithm is established; and the driving mode recognition model is trained, specifically including:

[0021] Initializing the network weights of the driving mode recognition model;

[0022] Extracting data from the effective feature vector affecting the driving mode according to a preset batch processing amount;

[0023] The Gaussian kernel function is used to map the input to a high-dimensional space, and the winning node is calculated according to the first formula; wherein the first formula is:

[0024]

[0025] In the formula, BMU represents the winning node; represents a nonlinear mapping from low-dimensional data to high-dimensional space; x i represents the i-th sample; v j represents the vector corresponding to the jth unit; k ii That is, K(x i ,x i ), K is the Gaussian kernel function; similarly, k ih That is, K(x i ,x h ); γ jh represents a binary function, when x h ∈j, γ jh =1, otherwise γ jh =0;

[0026] The network weights of the driving mode recognition model are updated according to the domain of the winning node.

[0027] As an improvement of the above solution, the driving mode recognition model is verified as follows:

[0028] The driving mode recognition model is verified by using a cross-validation method until the mean absolute error of the driving mode recognition model is less than or equal to a preset threshold.

[0029] As an improvement of the above solution, after training and verifying the driving mode recognition model, the method further includes:

[0030] Upload the trained driving mode recognition model to the vehicle-side edge computing platform or monitoring platform to obtain driving mode recognition results in real time.

[0031] An embodiment of the present invention further provides a driving mode recognition device, comprising:

[0032] A data acquisition module, used to acquire driver data, vehicle data and environmental data during vehicle driving; wherein the driver data includes driver status data and driver behavior data, the vehicle data includes vehicle operating condition data, battery data and time data, and the environmental data includes climate data and geographic data;

[0033] A preprocessing module, used for performing data preprocessing on the driver data, the vehicle data and the environment data;

[0034] The selection module is used to extract and analyze the preprocessed data through feature engineering technology, and select the effective feature vectors that affect the driving mode using PCA dimensionality reduction technology;

[0035] The recognition module is used to establish a driving mode recognition model according to the effective feature vector affecting the driving mode, and to train and verify the driving mode recognition model to obtain a driving mode recognition result.

[0036] An embodiment of the present invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements any of the above-mentioned driving mode recognition methods when executing the computer program.

[0037] An embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the driving mode recognition methods described above.

[0038] Compared with the prior art, the driving mode recognition method, device, terminal device and storage medium provided by the embodiment of the present invention have the following beneficial effects: by acquiring the driver data, vehicle data and environmental data during the driving process of the vehicle; wherein the driver data includes the driver status data and the driver behavior data, the vehicle data includes the vehicle working condition data, the battery data and the time data, and the environmental data includes the climate data and the geographical data; performing data preprocessing on the driver data, the vehicle data and the environmental data; extracting and analyzing the preprocessed data by feature engineering technology, and selecting the effective feature vectors that affect the driving mode by using the PCA dimensionality reduction technology; establishing a driving mode recognition model according to the effective feature vectors that affect the driving mode, and training and verifying the driving mode recognition model to obtain the driving mode recognition result. The embodiment of the present invention is based on a large amount of data during the driving process of the vehicle, and can accurately reflect the actual situation of the vehicle driving process. At the same time, the improved SOM algorithm is used to identify the driving mode, which can effectively improve the accuracy of driving mode recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of a preferred embodiment of a driving mode recognition method provided by the present invention;

[0040] Figure 2It is a structural schematic diagram of a preferred embodiment of a driving mode recognition device provided by the present invention;

[0041] Figure 3 It is a structural schematic diagram of a preferred embodiment of a terminal device provided by the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] See also Figure 1 , Figure 1 1 is a flow chart of a preferred embodiment of a driving mode recognition method provided by the present invention. The driving mode recognition method comprises:

[0044] S1, acquiring driver data, vehicle data and environmental data during vehicle driving; wherein the driver data includes driver status data and driver behavior data, the vehicle data includes vehicle operating condition data, battery data and time data, and the environmental data includes climate data and geographic data;

[0045] S2, performing data preprocessing on the driver data, the vehicle data and the environment data;

[0046] S3, feature extraction and analysis of the preprocessed data are performed through feature engineering technology, and PCA dimensionality reduction technology is used to select effective feature vectors that affect driving modes;

[0047] S4, establishing a driving mode recognition model according to the effective feature vector affecting the driving mode, and training and verifying the driving mode recognition model to obtain a driving mode recognition result.

[0048] Specifically, this embodiment first obtains the driver data, vehicle data and environmental data during the driving process of the vehicle; wherein the driver data includes the driver status data and the driver behavior data, the vehicle data includes the vehicle operating condition data, the battery data and the time data, and the environmental data includes the climate data and the geographical data. Then, the acquired driver data, vehicle data and environmental data are preprocessed to improve the data quality. Secondly, the preprocessed data is feature extracted and analyzed by feature engineering technology, and the PCA dimensionality reduction technology is used to select the effective feature vectors that affect the driving mode. Finally, a driving mode recognition model is established based on the effective feature vectors that affect the driving mode, and the driving mode recognition model is trained and verified to obtain the driving mode recognition result.

[0049] This embodiment is driven by human-vehicle-environment big data, slices the human-vehicle-environment data in different scenarios, uses PCA dimensionality reduction technology to select feature vectors that affect driving modes, and uses algorithms to perform machine learning on massive human-vehicle-environment historical data on a cloud computing platform, trains and verifies the model, and then obtains driving mode recognition results, effectively improving the accuracy of driving mode recognition, so that the driver can be reminded of safe and standardized operations based on the driving mode recognition results, providing the driver with personalized driving assistance.

[0050] In another preferred embodiment, the driver status data includes fatigue level, age and gender; the driver behavior data includes throttle data, braking data and sharp turn data; the vehicle operating condition data includes average vehicle speed, vehicle speed extremes, tire pressure and tire temperature; the battery data includes current, voltage and battery temperature; the time data includes single driving time and timestamp; the climate data includes temperature, humidity, season and light; the geographic data includes location, road conditions and terrain.

[0051] Specifically, this embodiment is different from the traditional driving mode recognition method with a small amount of data. Instead, it is calculated based on a number of data under seven major categories, including driver status, driver behavior, vehicle operating data, battery, time, climate and geography. It can accurately reflect the actual situation of the vehicle driving process, thereby improving the accuracy of driving mode recognition.

[0052] It should be noted that the specific data under the seven categories in this embodiment are not limited to the above data, as long as it is data useful for driving mode recognition.

[0053] In another preferred embodiment, the step S1, acquiring driver data, vehicle data and environment data during vehicle driving, specifically includes:

[0054] S101, obtaining VDCM domain data, BDCM domain data, and IDCM domain data of the vehicle through a big data platform; wherein VDCM refers to signals related to vehicle chassis power, BDCM refers to signals related to the vehicle body, and IDCM refers to signals related to the smart cockpit;

[0055] S102, screening the VDCM domain data, the BDCM domain data and the IDCM domain data to obtain the driver data, the vehicle data and the environment data during vehicle driving.

[0056] Specifically, this embodiment first determines the signals related to driving mode recognition, such as driver status, driver behavior, vehicle operating data, battery, time, climate and geography, and then collects data associated with the above signals from different data sources under the big data platform, and divides the above signals into time type and periodic type for collection, wherein the data source includes the VDCM domain related to vehicle chassis power, the BDCM domain related to the vehicle body and the IDCM domain related to the smart cockpit. Then, the data obtained from the above data sources are screened to obtain the driver data, vehicle data and environmental data during the vehicle driving process.

[0057] In another preferred embodiment, the step S2 of preprocessing the driver data, the vehicle data and the environment data comprises at least one of the following:

[0058] removing noise data and outliers from the driver data, the vehicle data, and the environment data;

[0059] For missing values ​​and invalid values ​​in the driver data, the vehicle data and the environment data, the adjacent average values ​​are selected for filling;

[0060] Performing logarithmic transformation on the relatively scattered data among the driver data, the vehicle data and the environment data;

[0061] Performing binning processing on the data according to data characteristics of the driver data, the vehicle data, and the environment data;

[0062] The driver data, the vehicle data and the environment data are standardized and normalized.

[0063] Specifically, when preprocessing the driver data, vehicle data, and environment data, one or more of the following methods may be used:

[0064] (1) Remove noise data and outliers from driver data, vehicle data, and environmental data;

[0065] (2) For missing values ​​and invalid values ​​in driver data, vehicle data, and environmental data, the adjacent average value is used to fill them;

[0066] (3) Perform logarithmic transformation on the scattered data among driver data, vehicle data and environmental data;

[0067] (4) binning the data according to the data characteristics of driver data, vehicle data, and environmental data;

[0068] (5) Standardize and normalize driver data, vehicle data, and environmental data.

[0069] In another preferred embodiment, establishing a driving mode recognition model specifically includes: establishing a driving mode recognition model based on an improved SOM algorithm; and training the driving mode recognition model specifically includes:

[0070] Initializing the network weights of the driving mode recognition model;

[0071] Extracting data from the effective feature vector affecting the driving mode according to a preset batch processing amount;

[0072] The Gaussian kernel function is used to map the input to a high-dimensional space, and the winning node is calculated according to the first formula; wherein the first formula is:

[0073]

[0074] In the formula, BMU represents the winning node; represents a nonlinear mapping from low-dimensional data to high-dimensional space; x i represents the i-th sample; v j represents the vector corresponding to the jth unit; k ii That is, K(x i ,x i ), K is the Gaussian kernel function; similarly, k ih That is, K(x i ,x h ); γ jh represents a binary function, when x h ∈j, γ jh =1, otherwise γ jh =0;

[0075] The network weights of the driving mode recognition model are updated according to the domain of the winning node.

[0076] Specifically, this embodiment establishes a driving mode recognition model based on the improved SOM algorithm according to the effective feature vectors that affect the driving mode, and divides the selected effective feature vectors that affect the driving mode into a training set, a test set, and a validation set according to a preset ratio, for example, 60% training set, 20% test set, and 20% validation set, and first trains the driving mode recognition model with the training set. Initialize the network weights of the driving mode recognition model; extract data from the effective feature vectors that affect the driving mode according to the preset batch processing volume; use the Gaussian kernel function to map the input to a high-dimensional space, and calculate the winning node according to the first formula; wherein the first formula is:

[0077]

[0078] In the formula, BMU represents the winning node; represents a nonlinear mapping from low-dimensional data to high-dimensional space; x i represents the i-th sample; v j represents the vector corresponding to the jth unit; k ii That is, K(x i ,x i ), K is the Gaussian kernel function; similarly, k ih That is, K(x i ,x h ); γ jh represents a binary function, when x h ∈j, γ jh =1, otherwise γ jh =0; Update the network weights of the driving mode recognition model according to the domain of the winning node.

[0079] Since the calculation of the classic SOM model is relatively complex and the training convergence time is slow, this embodiment uses SOM+ to add a scalable batch processing module. When performing quasi-real-time calculations, small batches are used for rapid calculations. In offline processing scenarios, the batch size is increased to make the model have better effects in both online and offline scenarios. At the same time, since there may be complex relationships between different driving modes, this method is less sensitive to noise than other clustering methods with faster convergence speeds. Even if there is a complex relationship between two clusters, SOM+ can better distinguish the two clusters than other methods. In addition, considering that the recognition results of the driving mode will eventually be used by other modules, SOM+ itself can retain the characteristics of the data topology structure and provide more information for subsequent users.

[0080] This embodiment is driven by human-vehicle-environment big data, slices the human-vehicle-environment data in different scenarios, uses PCA dimensionality reduction technology to select feature vectors that affect driving modes, and uses an improved SOM algorithm (SOM+) to perform machine learning on massive human-vehicle-environment historical data on a cloud computing platform, trains and verifies the model, and then obtains the driving mode recognition result, which effectively improves the accuracy of driving mode recognition, so that the driver can be reminded to operate safely and in accordance with the driving mode recognition result, and personalized driving assistance is provided to the driver.

[0081] In another preferred embodiment, the verifying the driving mode recognition model is specifically:

[0082] The driving mode recognition model is verified by using a cross-validation method until the mean absolute error of the driving mode recognition model is less than or equal to a preset threshold.

[0083] Specifically, the driving mode recognition model is cross-validated using a test set and a validation set, and the cross-validation is repeated k times, each time a test set or a validation set is selected, and the mean absolute error of the k times is compared with a preset threshold until the mean absolute error is less than or equal to the preset threshold.

[0084] As a preferred solution, after training and verifying the driving mode recognition model, the method further includes:

[0085] Upload the trained driving mode recognition model to the vehicle-side edge computing platform or monitoring platform to obtain driving mode recognition results in real time.

[0086] Specifically, after training and verifying the driving mode recognition model, a trained driving mode recognition model can be obtained. The driving mode recognition result can be obtained by inputting the test set into the driving mode recognition model. The trained driving mode recognition model can be uploaded to the vehicle-side edge computing platform or monitoring platform to obtain the driving mode recognition result in real time, and the driving mode can be fed back to the user and the background to remind the driver to operate safely and in accordance with regulations at all times.

[0087] Correspondingly, the present invention also provides a driving mode recognition device, which can implement all the processes of the driving mode recognition method in the above embodiment.

[0088] See also Figure 2 , Figure 2 1 is a schematic diagram of a preferred embodiment of a driving mode recognition device provided by the present invention. The driving mode recognition device comprises:

[0089] The data acquisition module 201 is used to acquire the driver data, vehicle data and environmental data during the driving process of the vehicle; wherein the driver data includes the driver status data and the driver behavior data, the vehicle data includes the vehicle operating condition data, the battery data and the time data, and the environmental data includes the climate data and the geographical data;

[0090] A preprocessing module 202, configured to perform data preprocessing on the driver data, the vehicle data and the environment data;

[0091] A selection module 203 is used to extract and analyze the preprocessed data by feature engineering technology, and select effective feature vectors that affect the driving mode by using PCA dimensionality reduction technology;

[0092] The recognition module 204 is used to establish a driving mode recognition model according to the effective feature vector affecting the driving mode, and to train and verify the driving mode recognition model to obtain a driving mode recognition result.

[0093] Preferably, the driver status data includes fatigue level, age and gender; the driver behavior data includes throttle data, braking data and sharp turn data; the vehicle operating condition data includes average vehicle speed, vehicle speed extremes, tire pressure and tire temperature; the battery data includes current, voltage and battery temperature; the time data includes single driving time and timestamp; the climate data includes temperature, humidity, season and light; the geographic data includes location, road conditions and terrain.

[0094] Preferably, the data acquisition module 201 is specifically used for:

[0095] Acquire VDCM domain data, BDCM domain data and IDCM domain data of the vehicle through the big data platform; wherein VDCM refers to signals related to vehicle chassis power, BDCM refers to signals related to the vehicle body, and IDCM refers to signals related to the smart cockpit;

[0096] The VDCM domain data, the BDCM domain data and the IDCM domain data are screened to obtain the driver data, the vehicle data and the environment data during vehicle driving.

[0097] Preferably, the preprocessing module 202 performs data preprocessing on the driver data, the vehicle data and the environment data, including at least one of the following:

[0098] removing noise data and outliers from the driver data, the vehicle data, and the environment data;

[0099] For missing values ​​and invalid values ​​in the driver data, the vehicle data and the environment data, the adjacent average values ​​are selected for filling;

[0100] Performing logarithmic transformation on the relatively scattered data among the driver data, the vehicle data and the environment data;

[0101] Performing binning processing on the data according to data characteristics of the driver data, the vehicle data, and the environment data;

[0102] The driver data, the vehicle data and the environment data are standardized and normalized.

[0103] Preferably, establishing a driving mode recognition model comprises: establishing a driving mode recognition model based on an improved SOM algorithm; and training the driving mode recognition model comprises:

[0104] Initializing the network weights of the driving mode recognition model;

[0105] Extracting data from the effective feature vector affecting the driving mode according to a preset batch processing amount;

[0106] The Gaussian kernel function is used to map the input to a high-dimensional space, and the winning node is calculated according to the first formula; wherein the first formula is:

[0107]

[0108] In the formula, BMU represents the winning node; In the formula, BMU represents the winning node; represents a nonlinear mapping from low-dimensional data to high-dimensional space; x i represents the i-th sample; v j represents the vector corresponding to the jth unit; k ii That is, K(x i ,x i ), K is the Gaussian kernel function; similarly, k ih That is, K(x i ,x h ); γ jh represents a binary function, when x h ∈j, γ jh =1, otherwise γ jh =0;

[0109] The network weights of the driving mode recognition model are updated according to the domain of the winning node.

[0110] Preferably, the verifying the driving mode recognition model is specifically:

[0111] The driving mode recognition model is verified by using a cross-validation method until the mean absolute error of the driving mode recognition model is less than or equal to a preset threshold.

[0112] Preferably, after training and verifying the driving mode recognition model, the recognition module 204 is further used to:

[0113] Upload the trained driving mode recognition model to the vehicle-side edge computing platform or monitoring platform to obtain driving mode recognition results in real time.

[0114] In specific implementation, the working principle, control process and technical effect achieved by the driving mode recognition device provided in the embodiment of the present invention are the same as those of the driving mode recognition method in the above embodiment, and will not be repeated here.

[0115] See also Figure 3 , Figure 3 3 is a schematic diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 301, a memory 302, and a computer program stored in the memory 302 and configured to be executed by the processor 301. When the processor 301 executes the computer program, the driving mode recognition method described in any of the above embodiments is implemented.

[0116] Preferably, the computer program can be divided into one or more modules / units (such as computer program 1, computer program 2, ...), and the one or more modules / units are stored in the memory 302 and executed by the processor 301 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0117] The processor 301 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor, or the processor 301 can be any conventional processor. The processor 301 is the control center of the terminal device, and various parts of the terminal device are connected using various interfaces and lines.

[0118] The memory 302 mainly includes a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc., and the data storage area can store related data, etc. In addition, the memory 302 can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, and a flash card (Flash Card), etc., or the memory 302 can also be other volatile solid-state storage devices.

[0119] It should be noted that the above terminal device may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that Figure 3 The structural diagram is only an example of the above-mentioned terminal device and does not constitute a limitation of the above-mentioned terminal device. It may include more or less components than shown in the figure, or combine certain components, or different components.

[0120] An embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the driving mode recognition method described in any of the above embodiments.

[0121] The embodiment of the present invention provides a driving mode recognition method, device, terminal equipment and storage medium, by acquiring driver data, vehicle data and environmental data during vehicle driving; wherein the driver data includes driver status data and driver behavior data, the vehicle data includes vehicle operating condition data, battery data and time data, and the environmental data includes climate data and geographic data; data preprocessing is performed on the driver data, the vehicle data and the environmental data; feature extraction and analysis are performed on the preprocessed data through feature engineering technology, and effective feature vectors affecting the driving mode are selected by using PCA dimensionality reduction technology; a driving mode recognition model based on an improved SOM algorithm is established according to the effective feature vector affecting the driving mode, and the driving mode recognition model is trained and verified to obtain a driving mode recognition result. The embodiment of the present invention is based on a large amount of data during vehicle driving, can accurately reflect the actual situation of the vehicle driving process, and at the same time, the improved SOM algorithm is used to identify the driving mode, which can effectively improve the accuracy of driving mode recognition.

[0122] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0123] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A driving mode recognition method, characterized in that: include: Acquiring driver data, vehicle data and environmental data during vehicle driving; wherein the driver data includes driver status data and driver behavior data, the vehicle data includes vehicle operating condition data, battery data and time data, and the environmental data includes climate data and geographic data; Preprocessing the driver data, the vehicle data and the environment data; extracting and analyzing features of the preprocessed data by feature engineering technology, and selecting effective feature vectors that affect the driving mode by PCA dimensionality reduction technology; Establishing a driving mode recognition model according to the effective feature vector affecting the driving mode, and training and verifying the driving mode recognition model to obtain a driving mode recognition result; The method for establishing a driving mode recognition model is specifically: establishing a driving mode recognition model based on an improved SOM algorithm; training the driving mode recognition model, specifically including: initializing the network weights of the driving mode recognition model; extracting data from the effective feature vector affecting the driving mode according to a preset batch processing amount; using a Gaussian kernel function to map the input to a high-dimensional space, and calculating a winning node according to a first formula; wherein the first formula is: In the formula, BMU represents the winning node; φ represents the nonlinear mapping from low-dimensional data to high-dimensional space; Represents the input vector x i The best matching unit of ; arg min means finding the parameter value that minimizes the following expression; υ φ represents the weight vector of the unit; x i represents the i-th sample; υ j represents the vector corresponding to the jth unit; k ii That is, k(x i ,x i ), k is the Gaussian kernel function, k ii is the inner product of the input vector itself calculated by the kernel function; similarly, k ih That is, k(x i ,x h ), k ih The input vector x i With x h The kernel function value of γ jh represents a binary function, γ jh For neurons j With input x h The association weight of jr : Neuron υ j With input x r The association weight of j,BMU : Neuron υ j The association weight with the input BMU; k r，BMU : Input vector x r and the kernel function value of BMU; The network weights of the driving mode recognition model are updated according to the domain of the winning node.

2. The driving mode recognition method according to claim 1, characterized in that: The driver status data includes fatigue level, age and gender; the driver behavior data includes throttle data, braking data and sharp turn data; the vehicle operating condition data includes average vehicle speed, vehicle speed extremes, tire pressure and tire temperature; the battery data includes current, voltage and battery temperature; the time data includes single driving time and timestamp; the climate data includes temperature, humidity, season and light; the geographic data includes location, road conditions and terrain.

3. The driving mode recognition method according to claim 2, characterized in that: The acquiring of the driver data, vehicle data and environmental data during vehicle driving specifically includes: acquiring the VDCM domain data, BDCM domain data and IDCM domain data of the vehicle through a big data platform; wherein, VDCM refers to signals related to vehicle chassis power, BDCM refers to signals related to the vehicle body, and IDCM refers to signals related to the smart cockpit; and filtering the VDCM domain data, BDCM domain data and IDCM domain data to obtain the driver data, vehicle data and environmental data during vehicle driving.

4. The driving mode recognition method according to claim 1, characterized in that: The data preprocessing of the driver data, the vehicle data and the environment data includes at least one of the following: removing noise data and outliers in the driver data, the vehicle data and the environment data; filling missing values ​​and invalid values ​​in the driver data, the vehicle data and the environment data with adjacent average values; performing logarithmic transformation on the scattered data in the driver data, the vehicle data and the environment data; Performing binning processing on the data according to data characteristics of the driver data, the vehicle data, and the environment data; The driver data, the vehicle data and the environment data are standardized and normalized.

5. The driving mode recognition method according to claim 4, characterized in that: The verifying the driving mode recognition model specifically includes: verifying the driving mode recognition model using a cross-validation method until the mean absolute error of the driving mode recognition model is less than or equal to a preset threshold.

6. The driving mode recognition method according to claim 1, characterized in that: After the driving mode recognition model is trained and verified, the method further includes: uploading the trained driving mode recognition model to a vehicle-side edge computing platform or a monitoring platform to obtain a driving mode recognition result in real time.

7. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the driving mode recognition method according to any one of claims 1 to 6 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the driving mode recognition method according to any one of claims 1 to 6.

Images