Driving style prediction method and system, vehicle and storage medium

Abstract

The invention provides a driving style prediction method and system, a vehicle and a storage medium. The method comprises the following steps: acquiring driving data in a vehicle driving process; preprocessing the driving data to obtain a training set, a verification set and a test set; sample data in the training set and the verification set are used for constructing and training a core prediction model and a reference prediction model; inputting the sample data in the test set into the core prediction model to obtain a core driving style; inputting the sample data in the test set into a reference prediction model to obtain a reference driving style; the target driving style is determined and output according to the core driving style and the reference driving style, so that the probability of misjudgment of the core prediction model during driving style prediction can be reduced, and the reliability of the prediction result of the driving style is improved.

Description

Technical Field

[0001] The present application relates to the technical field of driving style prediction, and in particular to a driving style prediction method, system, vehicle, and storage medium. Background Art

[0002] With advances in technology and the development of automotive electronics, modern vehicles are now widely equipped with automated driving features, providing drivers with a more convenient and relaxing driving experience, making road travel safer and more efficient. To tailor driving styles to user habits, related technologies typically determine driving styles based on vehicle parameters during driving, which suffers from low accuracy and poor generalization. Summary of the Invention

[0003] The main purpose of this application is to provide a driving style prediction method, system, vehicle and storage medium, aiming to solve the problem of how to improve the prediction accuracy in the driving style prediction process in the prior art.

[0004] The present application is introduced below from different aspects. It should be understood that the implementation methods and beneficial effects of the following different aspects can be referenced to each other.

[0005] In a first aspect, the present application provides a driving style prediction method, which is applied to an electronic device. The driving style prediction method includes: Preprocess the driving data to obtain a training set, a validation set, and a test set. The sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model. Input the sample data in the test set into the core prediction model to obtain the core driving style; Input the sample data in the test set into the reference prediction model to obtain the reference driving style; A target driving style is determined and output based on the core driving style and the reference driving style.

[0006] In conjunction with the first aspect, in one possible implementation, the driving style prediction method further includes: When it is determined that the target driving style is inconsistent with the subjective evaluation style input by the user, obtaining driving data corresponding to the subjective evaluation style; The driving data corresponding to the subjective evaluation style is preprocessed to update the training set, validation set, and test set.

[0007] In conjunction with the first aspect, in one possible implementation, determining and outputting a target driving style based on the core driving style and the reference driving style includes: When the core driving style is inconsistent with the reference driving style, a corresponding verification value is obtained according to the core driving style and the reference driving style; Determine the target driving style based on the obtained verification value and output it; When the core driving style is consistent with the reference driving style, the core driving style or the reference driving style is determined as the target driving style and output.

[0008] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the first driving style and the reference driving style is the second driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the second driving style as a primary verification value, and determining the probability value of the third driving style as a secondary verification value; The target driving style is determined based on the obtained calibration value and output, including: When the primary check value is greater than or equal to the first primary threshold and the secondary check value is greater than or equal to the first secondary threshold, determining the third driving style as the target driving style and outputting the style; When the primary check value is greater than or equal to the first primary threshold and the secondary check value is less than the first secondary threshold, determining the second driving style as the target driving style and outputting the same; When the primary check value is less than the first primary threshold and the secondary check value is greater than or equal to the second secondary threshold, determining the third driving style as the target driving style and outputting the third driving style; When the primary check value is less than the first primary threshold, the secondary check value is less than the second secondary threshold, and the secondary check value is greater than or equal to the third secondary threshold, determining the second driving style as the target driving style and outputting the result; When the primary check value is less than the first primary threshold and the secondary check value is less than the third secondary threshold, determining the core driving style as the target driving style and outputting the target driving style; The second secondary threshold is greater than the first secondary threshold, and the third secondary threshold is less than the second secondary threshold.

[0009] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the first driving style and the reference driving style is the third driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the first driving style as a primary verification value, and determining the probability value of the third driving style as a secondary verification value; The target driving style is determined based on the obtained calibration value and output, including: When the primary check value is less than or equal to the second primary threshold and the secondary check value is greater than or equal to the fourth secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is less than or equal to the second primary threshold and the secondary check value is less than a fourth secondary threshold, determining the second driving style as the target driving style and outputting the target driving style; When the main verification value is greater than the second main threshold, the core driving style is determined as the target driving style and output.

[0010] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the second driving style and the reference driving style is the first driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the first driving style as a primary verification value, and determining the probability value of the third driving style as a secondary verification value; The target driving style is determined based on the obtained calibration value and output, including: When the main check value is greater than or equal to a third main threshold, determining the first driving style as a target driving style and outputting the same; When the primary check value is less than the third primary threshold and the secondary check value is greater than or equal to the fifth secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is smaller than the third primary threshold and the secondary check value is smaller than the fifth secondary threshold, the core driving style is determined as the target driving style and output.

[0011] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value, a secondary check value, and a final check value; When the core driving style is the second driving style and the reference driving style is the third driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the third driving style as a primary verification value, determining the probability value of the first driving style as a secondary verification value, and determining the probability value of the second driving style as a final verification value; The target driving style is determined based on the obtained calibration value and output, including: When the primary check value is greater than or equal to a fourth primary threshold value and the secondary check value is greater than or equal to a sixth secondary threshold value, determining the first driving style as a target driving style and outputting the target driving style; When the primary check value is less than a fourth primary threshold and the secondary check value is less than a sixth secondary threshold, determining the third driving style as the target driving style and outputting the same; When the main check value is less than the fourth main threshold value and the final check value is less than the first final threshold value, determining the third driving style as the target driving style and outputting the same; and When the main check value is less than the fourth main threshold and the final check value is greater than or equal to the first final threshold, the core driving style is determined as the target driving style and output.

[0012] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the third driving style and the reference driving style is the first driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the third driving style as a primary verification value and determining the probability value of the second driving style as a secondary verification value; The target driving style is determined based on the obtained calibration value and output, including: When the primary check value is less than or equal to the fifth primary threshold and the secondary check value is greater than or equal to the seventh secondary threshold, determining the second driving style as the target driving style and outputting the same; When the primary check value is less than or equal to a fifth primary threshold and the secondary check value is less than a seventh secondary threshold, determining the first driving style as a target driving style and outputting the target driving style; and When the main verification value is greater than the fifth main threshold, the core driving style is determined as the target driving style and output.

[0013] In conjunction with the first aspect, in one possible implementation, the core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the third driving style and the reference driving style is the second driving style, obtaining corresponding verification values ​​according to the core driving style and the reference driving style includes: Determining the probability value of the second driving style as a primary verification value, and determining the probability value of the first driving style as a secondary verification value; The target driving style is determined based on the obtained calibration value and output, including: When the primary check value is greater than or equal to a sixth primary threshold and the secondary check value is greater than or equal to an eighth secondary threshold, determining the first driving style as the target driving style; determining the second driving style as the target driving style when the primary check value is greater than or equal to a sixth primary threshold and when the secondary check value is less than an eighth secondary threshold; and When the primary check value is less than a sixth primary threshold value and the secondary check value is greater than or equal to a ninth secondary threshold value, determining the second driving style as the target driving style; and When the primary check value is less than the sixth primary threshold and the secondary threshold is less than or equal to the ninth secondary threshold, determining the core driving style as the target driving style; The ninth secondary threshold is smaller than the eighth secondary threshold.

[0014] In a second aspect, the present application provides a driving style prediction system, the driving style prediction system comprising: A data acquisition module is used to obtain driving data of the vehicle during driving; A data processing module is used to pre-process the driving data to obtain a training set, a validation set, and a test set; wherein the sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model; Prediction module, used to: Input the sample data in the test set into the core prediction model to obtain the core driving style; Input the sample data in the test set into the reference prediction model to obtain the reference driving style; A target driving style is determined and output based on the core driving style and the reference driving style.

[0015] In a third aspect, the present application provides a vehicle, comprising a processor and a memory, wherein the memory is configured to store a plurality of program instructions, and when the processor calls the program instructions, the following steps are implemented: Preprocess the driving data to obtain a training set, a validation set, and a test set. The sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model. Input the sample data in the test set into the core prediction model to obtain the core driving style; Input the sample data in the test set into the reference prediction model to obtain the reference driving style; A target driving style is determined and output based on the core driving style and the reference driving style.

[0016] In a fourth aspect, the present application provides a computer-readable storage medium storing computer-executable instructions that can be executed by at least one processor. When the computer-executable instructions are executed by at least one processor, the following steps are implemented: Preprocess the driving data to obtain a training set, a validation set, and a test set. The sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model. Input the sample data in the test set into the core prediction model to obtain the core driving style; Input the sample data in the test set into the reference prediction model to obtain the reference driving style; A target driving style is determined and output based on the core driving style and the reference driving style.

[0017] Compared with the prior art, this application has the following advantages: In the embodiments of this application, a core prediction model and a reference prediction model are used to analyze driving data from two dimensions, respectively. A target driving style is determined based on the core and reference driving styles. This reduces the probability of misjudgments by the core prediction model when predicting driving styles, thereby improving the reliability of driving style prediction results. Furthermore, the combination of the core and reference prediction models to optimize prediction results reduces the complexity of building the core prediction model compared to optimizing the core prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without any creative work.

[0019] Figure 1 This is a flowchart of the driving style prediction method provided in an embodiment of the present application.

[0020] Figure 2 for Figure 1 Detailed flowchart of step S14 in FIG.

[0021] Figure 3 for Figure 2 A detailed flowchart of step S143 of the first embodiment.

[0022] Figure 4 for Figure 2 A detailed flowchart of step S143 of the second embodiment.

[0023] Figure 5 for Figure 2 A detailed flowchart of step S143 of the third embodiment.

[0024] Figure 6 for Figure 2A detailed flowchart of step S143 of the fourth embodiment.

[0025] Figure 7 for Figure 2 A detailed flowchart of step S143 of the fifth embodiment.

[0026] Figure 8 for Figure 2 A detailed flowchart of step S143 of the sixth embodiment.

[0027] Figure 9 A schematic diagram of the modules of the driving style prediction system provided in an embodiment of the present application.

[0028] Figure 10 A schematic diagram of the modules of a vehicle provided in an embodiment of the present application.

[0029] Description of main component symbols Steps S10-S16, S141-S144, S1431-S1437 Vehicle 100 Driving style prediction system 1 Sensor 2 Vehicle Controller 3 Input / output devices 4 Processor 5 Memory 6 Communication interface 7 Data acquisition module 10 Data processing module 20 Prediction module 30 Verification Module 40 First prediction unit 31 Second prediction unit 32 Result detection unit 33 The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings 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 in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0031] The terms "first", "second", and "third" in the specification of this application and the above drawings are used to distinguish different objects rather than to describe a specific order. In addition, the term "including" and any variations thereof are intended to cover non-exclusive inclusions.

[0032] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections, electrical connections, or mutual communication; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to internal communication between two components or the interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on specific circumstances.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein in the specification of this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0034] See also Figure 1 , which is a flow chart of a driving style prediction method according to a preferred embodiment of the present application. The driving style prediction method can be applied to a terminal, a server, or a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In at least one embodiment of the present application, the terminal can be a vehicle 100 (e.g., Figure 4 As shown in FIG. 1 , the vehicle 100 may include a driving style prediction system 1 , a sensor 2 , a vehicle controller 3 , and an input / output device 4 .

[0035] Step S10: Acquire driving data of the vehicle during driving.

[0036] In at least one embodiment of the present application, driving data includes vehicle status data and vehicle control data. Vehicle status data includes vehicle wheel speed, vehicle speed, vehicle acceleration, vehicle yaw rate, distance between the vehicle and the preceding vehicle, vehicle drive torque, etc. Vehicle control data includes steering wheel angle / speed, brake pedal depth, accelerator pedal depth, vehicle latitude and longitude coordinates, etc. Driving data can be obtained by sensor 2 and vehicle controller 3.

[0037] Step S11: pre-process the driving data to obtain a training set, a validation set, and a test set.

[0038] To increase data processing speed, the driving data may be preprocessed. In at least one embodiment of the present application, the preprocessing of the driving data includes steps such as segment cropping, operating condition labeling, driving style labeling, feature extraction, normalization, and sample partitioning.

[0039] The segmented clipping operation is mainly used to remove invalid and abnormal data in the driving data and obtain multiple data segments, such as data with obvious glitches, data in a stationary state of the vehicle, and data during a vehicle driving at a constant speed.

[0040] The operating condition labeling operation primarily identifies the operating condition type corresponding to each data segment. In at least one embodiment of the present application, operating condition types include turning / U-turning, overtaking / lane changing, starting / stopping, and accelerating / decelerating. In other embodiments, the operating condition types may be further or less diverse as needed. The operating condition labeling is based on identifying the operating condition every 5 seconds within a 10-second time period. For example, when the sum of the squared yaw rates exceeds 80, the absolute value of the vehicle's steering angle is greater than 30 degrees, or the absolute value of the maximum yaw rate exceeds 7.5 degrees, and the average vehicle speed remains between 15 and 60, a turning / U-turn condition is determined. If the sum of the squared yaw rates exceeds 200, the absolute value of the vehicle's steering angle is greater than 2.5 degrees, or the sum of the squared yaw rates exceeds 150, the absolute value of the vehicle's steering angle is greater than 3.5 degrees, the absolute value of the vehicle's steering angle is less than 30 degrees, the average vehicle speed exceeds 60, and the minimum speed exceeds 50, a high-speed lane change / overtaking condition is determined. If the sum of the squared yaw rates is less than 150, or the absolute value of the vehicle's steering angle is less than 30 degrees, the minimum vehicle speed is less than 3, the maximum vehicle speed exceeds 10, and the accelerator pedal position is greater than 18, or the brake pedal position is greater than 20, a starting / stopping condition is determined. If the absolute value of the vehicle's steering angle is less than 30 degrees, the minimum speed is greater than 15, the average speed is greater than 25, and the absolute value of the maximum yaw angular velocity is less than 7.5 degrees, and the maximum speed is at least 18 greater than the minimum speed, then it is determined to be an acceleration / deceleration condition.

[0041] The driving style labeling operation is mainly used to label the evaluation data with a driving style using the latitude and longitude position information of the vehicle 100 .

[0042] The feature extraction operation is mainly used to identify the parameters that have an important impact on driving style among the vehicle state parameters, vehicle handling parameters, derived parameters (such as impact degree) and statistical characteristic parameters (including mean, variance, maximum value, etc.) of the data segment, and to screen out some parameters, such as vehicle state parameters, their derived parameters and statistical characteristic parameters, as input parameters of the machine learning model.

[0043] The normalization operation is mainly used to normalize the extracted feature data to obtain sample data.

[0044] The sample partitioning operation is mainly used to divide sample data into training, validation, and test sets. The training, validation, and test sets are used to improve the applicability and generalization ability of machine learning models in different usage scenarios.

[0045] Among them, the sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model.

[0046] In at least one embodiment of the present application, the core prediction model is a deep learning model. Specifically, a deep learning model cluster is constructed through a multi-stage training strategy, and the deep learning model with the best overall performance is screened based on cross-validation. After the core prediction model is constructed, the hyperparameters of the core prediction model are adjusted using the validation data in the validation set, so that the accuracy of the core model reaches a preset value. In at least one embodiment of the present application, the preset value can be 92%. Among them, the hyperparameters include learning rate, batch size, dropout value, optimizer selection, and weight decay coefficient.

[0047] In at least one embodiment of the present application, the core prediction model includes a convolutional layer, a long short-term memory layer, and an attention layer. The convolutional layer may be a convolutional neural network, the long short-term memory layer may be a long short-term memory neural network, and the attention layer may be a neural network with an attention mechanism. Specifically, the convolutional layer is used to perform convolution operations on the driving data in the training set to obtain global information about the driving data; the historical long short-term memory layer is used to determine the temporal delay of the driving data based on the global information of the driving data and output local information of the training set; the attention layer is used to output core driving style data based on the local information with time delay output by the long short-term memory layer. The core driving style data includes the core driving style and the corresponding probabilities of multiple driving styles. The types of multiple driving styles may include sporty, robust, and soft.

[0048] In at least one embodiment of the present application, the reference prediction model is a K nearest neighbor model. The K nearest neighbor model is a model constructed based on the K-Nearest Neighbors (KNN) algorithm. The K nearest neighbor model classifies data based on distance and can effectively distinguish different driving styles. At the same time, compared with the deep learning model, the K nearest neighbor model has a faster output speed. Specifically, the KNN algorithm is used to construct a K nearest neighbor model cluster based on the driving data corresponding to the training set, and based on the accuracy of the output data of the K nearest neighbor model under different K values ​​under cross-validation conditions, the K value corresponding to the highest accuracy is selected as the K value of the reference prediction model. At the same time, the voting weights of the K neighbor samples can be set during the verification process, and the closer the distance, the higher the corresponding weight value. Among them, the accuracy of the reference prediction model is greater than the second preset value. In at least one embodiment of the present application, the second preset value is 85%.

[0049] Step S12: inputting the sample data in the test set into the core prediction model to obtain the core driving style.

[0050] Step S13: input the sample data in the test set into the reference prediction model to obtain a reference driving style.

[0051] In at least one embodiment of the present application, the core driving style and the reference driving style are each at least one of a first driving style, a second driving style, and a third driving style. In at least one embodiment of the present application, the first driving style is a sporty driving style, the second driving style is a robust driving style, and the third driving style is a soft driving style.

[0052] Step S14: determining and outputting a target driving style based on the core driving style and the reference driving style.

[0053] The core prediction model may misjudge the distinction between robust and soft driving styles. To improve the accuracy of the prediction results, this application combines the core driving style with the reference driving style to determine the target driving style. This corrects the core driving style output by the deep learning model and improves the reliability of the driving style prediction results. After performing the correction operation, the accuracy of the core driving style is greater than a third preset value. The third preset value is greater than the first preset value and greater than the second preset value. In at least one embodiment of the present application, the third preset value is 95%.

[0054] Please also refer to Figure 2 , which is a detailed flowchart of step S14.

[0055] Step S141: Determine whether the core driving style is consistent with the reference driving style.

[0056] Step S142 : when the core driving style is inconsistent with the reference driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style.

[0057] In the first embodiment of the present application, when the core driving style is the first driving style and the reference driving style is the second driving style, the probability value of the second driving style is determined as the primary verification value, and the probability value of the third driving style is determined as the secondary verification value.

[0058] In a second embodiment of the present application, when the core driving style is the first driving style and the reference driving style is the third driving style, the probability value of the first driving style is determined as the primary verification value, and the probability value of the third driving style is determined as the secondary verification value.

[0059] In a third embodiment of the present application, when the core driving style is the second driving style and the reference driving style is the first driving style, the probability value of the first driving style is determined as the primary verification value, and the probability value of the third driving style is determined as the secondary verification value.

[0060] In a fourth embodiment of the present application, when the core driving style is the second driving style and the reference driving style is the third driving style, the probability value of the third driving style is determined as the primary verification value, the probability value of the second driving style is determined as the secondary verification value, and the probability value of the first driving style is determined as the final verification value.

[0061] In a fifth embodiment of the present application, when the core driving style is the third driving style and the reference driving style is the first driving style, the probability value of the third driving style is determined as the primary verification value, and the probability value of the second driving style is determined as the secondary verification value.

[0062] In a sixth embodiment of the present application, when the core driving style is the third driving style and the reference driving style is the second driving style, the probability value of the second driving style is determined as the primary verification value, and the probability value of the first driving style is determined as the secondary verification value.

[0063] Step S143: Determine the target driving style based on the obtained verification value and output it.

[0064] See also Figure 3 , which is a detailed flowchart of step S143 of the first embodiment. In the first embodiment, the core driving style is the first driving style and the reference driving style is the second driving style, the probability value of the second driving style is the primary check value, and the probability value of the third driving style is the secondary check value.

[0065] Step S1431, determine whether the primary check value is greater than or equal to the first primary threshold.

[0066] In at least one embodiment of the present application, the first main threshold is 7%.

[0067] Step S1432: When the primary check value is greater than or equal to the first primary threshold, determine whether the secondary check value is greater than or equal to the first secondary threshold.

[0068] In at least one embodiment of the present application, the first secondary threshold is 25%.

[0069] Step S1433: When the secondary check value is greater than or equal to the first secondary threshold, the third driving style is used as the target driving style.

[0070] In at least one embodiment of the present application, when the primary check value is greater than or equal to 7% and the secondary check value is greater than or equal to 25%, the third driving style corresponding to the secondary check value in the core prediction model is identified as being closer to the test data and the third driving style is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style.

[0071] Step S1434: When the secondary check value is less than the first secondary threshold, the second driving style is used as the target driving style.

[0072] In at least one embodiment of the present application, when the primary checksum is greater than or equal to 7% and the secondary checksum is less than 25%, the second driving style corresponding to the primary checksum in the core prediction model is identified as being closer to the test data and is used as the reference target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style.

[0073] Step S1435: When the primary check value is less than the first primary threshold, determine whether the secondary check value is greater than or equal to the second secondary threshold.

[0074] In at least one embodiment of the present application, the second secondary threshold is 40%.

[0075] When the secondary check value is greater than or equal to the second secondary threshold, proceed to step S1433.

[0076] In at least one embodiment of the present application, when the primary checksum is less than 7% and the secondary checksum is greater than or equal to 40%, the third driving style corresponding to the secondary checksum in the core prediction model is identified as being closer to the test data and is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style.

[0077] Step S1436: When the secondary check value is less than the second secondary threshold, determine whether the secondary check value is greater than or equal to the third secondary threshold.

[0078] In at least one embodiment of the present application, the third secondary threshold is 20%.

[0079] When the secondary check value is greater than or equal to the third secondary threshold, proceed to step S1434.

[0080] In at least one embodiment of the present application, when the primary checksum is less than 7% and the secondary checksum is less than 40% and greater than 20%, the second driving style corresponding to the primary checksum in the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style.

[0081] Step S1437: When the secondary check value is less than the third secondary threshold, the core driving style is used as the target driving style.

[0082] In at least one embodiment of the present application, when the primary checksum is less than 7% and the secondary checksum is less than 20%, the core driving style output by the core prediction model is determined to be close to the test data, and the first driving style output by the core prediction model is used as the target driving style. In other words, the core driving style output by the core prediction model is accurate, and no correction is required.

[0083] See also Figure 4 , which is a detailed flowchart of step S143 of the second embodiment. In the second embodiment, the core driving style is the first driving style and the reference driving style is the third driving style. The probability value of the first driving style is the primary check value and the probability value of the third driving style is the secondary check value.

[0084] Step S1431, determine whether the primary check value is less than or equal to the second primary threshold.

[0085] In at least one embodiment of the present application, the second main threshold is 95.3%.

[0086] Step S1432: When the primary check value is less than or equal to the second primary threshold, determine whether the secondary check value is greater than or equal to the fourth secondary threshold.

[0087] In at least one embodiment of the present application, the fourth secondary threshold is 45%.

[0088] Step S1433: When the secondary check value is greater than or equal to the fourth secondary threshold, the third driving style is used as the target driving style.

[0089] In at least one embodiment of the present application, when the primary checksum is less than or equal to 95.3% and the secondary checksum is greater than or equal to 45%, the third driving style corresponding to the secondary checksum in the core prediction model is identified as being closer to the test data and the third driving style is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style.

[0090] Step S1434: When the secondary check value is less than the fourth secondary threshold, the second driving style is used as the target driving style.

[0091] In at least one embodiment of the present application, when the primary checksum is less than or equal to 95.3% and the secondary checksum is less than 45%, the second driving style corresponding to the final checksum in the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style.

[0092] Step S1435: When the main verification value is greater than the second main threshold, the core driving style is used as the target driving style.

[0093] In at least one embodiment of the present application, when the primary checksum is less than 95.3%, the core driving style output by the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, the core driving style output by the core prediction model is correct and no correction is required.

[0094] See also Figure 5 , which is a detailed flowchart of step S143 of the third embodiment. In the third embodiment, the core driving style is the second driving style and the reference driving style is the first driving style. The probability value of the first driving style is the primary check value and the probability value of the third driving style is the secondary check value.

[0095] Step S1431, determine whether the primary check value is greater than or equal to the third primary threshold.

[0096] In at least one embodiment of the present application, the third main threshold is 40%.

[0097] In step S1432, when the primary check value is greater than or equal to the third primary threshold, the first driving style is used as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the first driving style.

[0098] In at least one embodiment of the present application, when the main check value is greater than or equal to 40%, the first driving style corresponding to the main check value in the core prediction model is identified as being closer to the test data, and the first driving style is used as the target driving style.

[0099] Step S1433: When the primary check value is less than the third primary threshold, determine whether the secondary check value is greater than or equal to the fifth secondary threshold.

[0100] In at least one embodiment of the present application, the fifth secondary threshold is 21.5%.

[0101] In step S1434, when the secondary check value is greater than or equal to the fifth secondary threshold, the third driving style is used as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style.

[0102] In at least one embodiment of the present application, when the primary check value is less than 40% and the secondary check value is greater than or equal to 21.5%, the third driving style corresponding to the secondary check value in the core prediction model is identified as being closer to the test data, and the third driving style is used as the target driving style.

[0103] Step S1435 : When the secondary check value is less than the fifth secondary threshold, the core driving style is used as the target driving style.

[0104] In at least one embodiment of the present application, when the primary checksum is less than 40% and the secondary checksum is less than 21.5%, the core driving style output by the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, the core driving style output by the core prediction model is correct, and no correction is required.

[0105] See also Figure 6 , which is a detailed flowchart of step S143 of the fourth embodiment. In the fourth embodiment, the core driving style is the second driving style and the reference driving style is the third driving style. The probability value of the third driving style is determined as the primary check value, the probability value of the second driving style is determined as the secondary check value, and the probability value of the first driving style is determined as the final check value.

[0106] Step S1431, determine whether the primary check value is greater than or equal to a fourth primary threshold.

[0107] In at least one embodiment of the present application, the fourth main threshold is 5%.

[0108] Step S1432: When the primary check value is greater than or equal to the fourth primary threshold, determine whether the secondary check value is greater than or equal to the sixth secondary threshold.

[0109] In at least one embodiment of the present application, the sixth secondary threshold is 29.8%.

[0110] Step S1433: When the secondary check value is greater than or equal to the sixth secondary threshold, the first driving style is used as the target driving style.

[0111] In at least one embodiment of the present application, when the primary check value is greater than or equal to 5% and the secondary check value is greater than or equal to 29.8%, the first driving style corresponding to the secondary check value in the core prediction model is identified as being closer to the test data and is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the first driving style.

[0112] Step S1434: When the secondary check value is less than the sixth secondary threshold, the third driving style is used as the target driving style.

[0113] In at least one embodiment of the present application, when the primary checksum is greater than or equal to 5% and the secondary checksum is less than 29.8%, the third driving style corresponding to the primary checksum in the core prediction model is identified as being closer to the test data and is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style.

[0114] Step S1435: When the main check value is less than the fourth main threshold, determine whether the final check value is less than the first final threshold.

[0115] In at least one embodiment of the present application, the first final-stage threshold is 77%.

[0116] When the final-stage check value is less than the first final-stage threshold, the process returns to step S1434.

[0117] In at least one embodiment of the present application, when the primary checksum is less than 5% and the final checksum is less than 77%, the third driving style corresponding to the secondary checksum in the core prediction model is identified as being closer to the test data and the third driving style is selected as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style.

[0118] Step S1436 : When the final-level verification value is greater than the first final-level threshold, the core driving style is used as the target driving style.

[0119] In at least one embodiment of the present application, when the primary checksum is less than 5% and the secondary checksum is greater than or equal to 77%, the core driving style output by the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, the core driving style output by the core prediction model is correct, and no correction is required.

[0120] See also Figure 7, which is a detailed flowchart of step S143 of the fifth embodiment. In the fifth embodiment, the core driving style is the third driving style and the reference driving style is the first driving style. The probability value of the third driving style is the primary check value and the probability value of the second driving style is the secondary check value.

[0121] Step S1431, determine whether the primary check value is less than or equal to the fifth primary threshold.

[0122] In at least one embodiment of the present application, the fifth main threshold is 55%.

[0123] Step S1432: When the primary check value is greater than or equal to the fifth primary threshold, determine whether the secondary check value is greater than or equal to the seventh secondary threshold.

[0124] In at least one embodiment of the present application, the seventh secondary threshold is 16%.

[0125] Step S1433: When the secondary check value is greater than or equal to the seventh secondary threshold, the second driving style is used as the target driving style.

[0126] In at least one embodiment of the present application, when the primary checksum is less than or equal to 55% and the secondary checksum is greater than or equal to 16%, the second driving style corresponding to the secondary checksum in the core prediction model is identified as being closer to the test data and the second driving style is used as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the second driving style.

[0127] Step S1434: When the secondary check value is less than the seventh secondary threshold, the first driving style is used as the target driving style.

[0128] In at least one embodiment of the present application, when the primary check value is greater than or equal to 55% and the secondary check value is less than 16%, the first driving style corresponding to the final check value in the core prediction model is identified as being closer to the test data, and the first driving style is used as the target driving style.

[0129] Step S1435 : When the main verification value is greater than or equal to the fifth main threshold, the core driving style is used as the target driving style.

[0130] In at least one embodiment of the present application, when the primary checksum is less than 5% and the secondary checksum is greater than or equal to 77%, the core driving style output by the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, the core driving style output by the core prediction model is correct, and no correction is required.

[0131] See also Figure 8, which is a detailed flowchart of step S143 of the sixth embodiment. In the sixth embodiment, the core driving style is the third driving style and the reference driving style is the second driving style. The probability value of the second driving style is the primary check value and the probability value of the first driving style is the secondary check value.

[0132] Step S1431, determine whether the primary check value is greater than or equal to the sixth primary threshold.

[0133] In at least one embodiment of the present application, the sixth main threshold is 21%.

[0134] Step S1432: When the primary check value is greater than or equal to the sixth primary threshold, determine whether the secondary check value is greater than or equal to the eighth secondary threshold.

[0135] In at least one embodiment of the present application, the eighth secondary threshold is 12.5%.

[0136] Step S1433: When the secondary check value is greater than or equal to the eighth secondary threshold, the first driving style is used as the target driving style.

[0137] In at least one embodiment of the present application, when the primary check value is greater than or equal to 21% and the secondary check value is greater than or equal to 12.5%, the first driving style corresponding to the secondary check value in the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the first driving style.

[0138] Step S1434: When the secondary check value is less than the eighth secondary threshold, the second driving style is used as the target driving style.

[0139] In at least one embodiment of the present application, when the primary checksum is greater than or equal to 21% and the secondary checksum is less than 12.5%, the second driving style corresponding to the primary checksum in the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the second driving style.

[0140] Step S1435: When the primary check value is less than the sixth primary threshold, determine whether the secondary check value is greater than or equal to the ninth secondary threshold.

[0141] In at least one embodiment of the present application, the ninth secondary threshold is 5%.

[0142] When the secondary check value is greater than or equal to the ninth secondary threshold, the process returns to step S1434.

[0143] Step S1436: When the secondary check value is less than the ninth secondary threshold, the third driving style is used as the target driving style.

[0144] In at least one embodiment of the present application, when the primary checksum is less than 21% and the secondary checksum is less than 5%, the core driving style output by the core prediction model is identified as being closer to the test data and is used as the target driving style. In other words, the core driving style output by the core prediction model is correct, and no correction is required.

[0145] Step S144 : When the core predicted driving style is consistent with the reference predicted driving style, the core predicted data or the reference predicted driving style is used as the final predicted driving style.

[0146] Step S15: determining whether the subjective evaluation style is consistent with the target driving style.

[0147] In at least one embodiment of the present application, the subjective data includes evaluation data provided by a user based on subjective judgment via the input / output device 4 of the vehicle 100. The evaluation data may be evaluation data provided by the user when the vehicle 100 passes through key locations (e.g., turns, roundabouts, gates) or travels on specific experimental sections (corresponding to different road types).

[0148] Step S16 : When the subjective evaluation style is inconsistent with the target driving style, the driving data corresponding to the subjective evaluation style is obtained, and the process returns to step S11 .

[0149] When the subjective driving style is consistent with the predicted driving style, the core prediction model is identified as meeting the requirements and the process ends.

[0150] Based on the aforementioned driving style prediction method, a core prediction model and a reference prediction model are used to analyze driving data from two different dimensions. The target driving style is determined by combining the core and reference driving styles. This reduces the probability of misjudgments by the core prediction model and improves the reliability of driving style prediction results. Furthermore, optimizing the prediction results by combining the core and reference prediction models reduces the complexity of building the core prediction model compared to optimizing the core prediction model.

[0151] See also Figure 9 , which is a module diagram of a driving style prediction system 1 according to at least one embodiment of the present application. The driving style prediction system 1 includes a data acquisition module 10 , a data processing module 20 , a prediction module 30 , and a verification module 40 .

[0152] The data acquisition module 10 is used to obtain driving data of the vehicle during driving.

[0153] In at least one embodiment of the present application, driving data includes vehicle status data and vehicle control data. Vehicle status data includes vehicle wheel speed, vehicle speed, vehicle acceleration, vehicle yaw rate, distance between the vehicle and the preceding vehicle, and vehicle drive torque. Vehicle control data includes steering wheel angle / speed, brake pedal depth, accelerator pedal depth, and vehicle latitude and longitude coordinates. Driving data can be obtained via sensor 2 and vehicle controller 3.

[0154] The data processing module 20 is used to pre-process the driving data to obtain a training set, a validation set, and a test set, and output sample data.

[0155] In order to improve the data processing speed, the driving data can be preprocessed. In at least one embodiment of the present application, the preprocessing operation of the driving data includes the steps of segmentation clipping, working condition labeling, driving style labeling, feature extraction, normalization, sample division, etc. The segmentation clipping operation is mainly used to remove invalid and abnormal data in the driving data and obtain multiple data segments, such as data with obvious glitches, data of the vehicle in a stationary state, and data of the vehicle in a uniform speed driving process. The working condition labeling operation is mainly used to label the working condition type corresponding to each data segment. In at least one embodiment of the present application, the working condition types include turning / U-turning, overtaking / changing lanes, starting / stopping, acceleration / deceleration. In other embodiments, the working condition types can be divided into more or less categories according to needs. Among them, the judgment basis for working condition labeling is to perform working condition identification every 5 seconds within a 10-second time period. For example, when the sum of the squared yaw rates exceeds 80, the absolute value of the vehicle's steering angle is greater than 30 degrees, or the absolute value of the maximum yaw rate exceeds 7.5 degrees, and the average vehicle speed remains between 15 and 60, a turning / U-turn condition is determined. If the sum of the squared yaw rates exceeds 200, the absolute value of the vehicle's steering angle is greater than 2.5 degrees, or the sum of the squared yaw rates exceeds 150, the absolute value of the vehicle's steering angle is greater than 3.5 degrees, the absolute value of the vehicle's steering angle is less than 30 degrees, the average vehicle speed exceeds 60, and the minimum speed exceeds 50, a high-speed lane change / overtaking condition is determined. If the sum of the squared yaw rates is less than 150, or the absolute value of the vehicle's steering angle is less than 30 degrees, the minimum vehicle speed is less than 3, the maximum vehicle speed exceeds 10, and the accelerator pedal position is greater than 18, or the brake pedal position is greater than 20, a starting / stopping condition is determined. If the absolute value of the vehicle's steering angle is less than 30 degrees, the minimum speed is greater than 15 degrees, the average speed is greater than 25 degrees, and the absolute value of the maximum yaw rate is less than 7.5 degrees, and the maximum speed is at least 18 degrees greater than the minimum speed, then the vehicle is considered to be in an acceleration / deceleration condition. The driving style labeling operation primarily uses the latitude and longitude location information of vehicle 100 to label the evaluation data with driving style. The feature extraction operation primarily identifies parameters within the data segment's vehicle state parameters, vehicle handling parameters, derived parameters (such as jerkiness), and statistical characteristic parameters (including mean, variance, and maximum) that have a significant impact on driving style. These parameters, such as vehicle state parameters, their derived parameters, and statistical characteristic parameters, are then selected as input parameters for the machine learning model. The normalization operation primarily normalizes the extracted feature data to obtain sample data. The sample partitioning operation primarily divides the sample data into training, validation, and test sets. The training, validation, and test sets are used to improve the applicability and generalization capabilities of the machine learning model in different usage scenarios.

[0156] Among them, the sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model.

[0157] In at least one embodiment of the present application, the core prediction model is a deep learning model. Specifically, a deep learning model cluster is constructed through a multi-stage training strategy, and the deep learning model with the best overall performance is screened based on cross-validation. After the core prediction model is constructed, the hyperparameters of the core prediction model are adjusted using the validation data in the validation set, so that the accuracy of the core model reaches a preset value. In at least one embodiment of the present application, the preset value can be 92%. Among them, the hyperparameters include learning rate, batch size, dropout value, optimizer selection, and weight decay coefficient.

[0158] In at least one embodiment of the present application, the core prediction model includes a convolutional layer, a long short-term memory layer, and an attention layer. The convolutional layer may be a convolutional neural network, the long short-term memory layer may be a long short-term memory neural network, and the attention layer may be a neural network with an attention mechanism. Specifically, the convolutional layer is used to perform convolution operations on the driving data in the training set to obtain global information about the driving data; the historical long short-term memory layer is used to determine the temporal delay of the driving data based on the global information of the driving data and output local information of the training set; the attention layer is used to output core driving style data based on the local information with time delay output by the long short-term memory layer. The core driving style data includes the core driving style and the corresponding probabilities of multiple driving styles. The types of multiple driving styles may include sporty, robust, and soft.

[0159] In at least one embodiment of the present application, the reference prediction model is a K nearest neighbor model. The K nearest neighbor model is a model constructed based on the K-Nearest Neighbors (KNN) algorithm. The K nearest neighbor model classifies data based on distance and can effectively distinguish different driving styles. At the same time, compared with the deep learning model, the K nearest neighbor model has a faster output speed. Specifically, the KNN algorithm is used to construct a K nearest neighbor model cluster based on the driving data corresponding to the training set, and based on the accuracy of the output data of the K nearest neighbor model under different K values ​​under cross-validation conditions, the K value corresponding to the highest accuracy is selected as the K value of the reference prediction model. At the same time, the voting weights of the K neighbor samples can be set during the verification process, and the closer the distance, the higher the corresponding weight value. Among them, the accuracy of the reference prediction model is greater than the second preset value. In at least one embodiment of the present application, the second preset value is 85%.

[0160] The prediction module 30 is configured to input the sample data in the test set into the core prediction model to obtain a core driving style. The prediction module 30 is also configured to input the sample data in the test set into the reference prediction model to obtain a reference driving style, and to determine and output a target driving style based on the core driving style and the reference driving style.

[0161] The prediction module 30 includes a first prediction unit 31 , a second prediction unit 32 , and a result detection unit 33 .

[0162] The first prediction unit 31 is electrically connected to the data processing module 20. The first prediction unit 31 is configured to construct and train a core prediction model based on sample data from the training and validation sets. The first prediction unit 31 is also configured to input sample data from the test set into the core prediction model to determine the core driving style.

[0163] In at least one embodiment of the present application, the core prediction model is a deep learning model. Specifically, a deep learning model cluster is constructed through a multi-stage training strategy, and the deep learning model with the best overall performance is screened based on cross-validation. After the core prediction model is constructed, the hyperparameters of the Hexing prediction model are adjusted using the validation data in the validation set, so that the accuracy of the core model reaches a preset value. In at least one embodiment of the present application, the preset value can be 92%. Among them, the hyperparameters include the learning rate, batch size, dropout value, optimizer selection, and weight decay coefficient.

[0164] The second prediction unit 32 is electrically connected to the data processing module 20. The second prediction unit 32 is used to construct and train a reference prediction model using sample data from the training and validation sets. The second prediction unit 32 is also used to input sample data from the test set into the reference prediction model to obtain a reference driving style.

[0165] In at least one embodiment of the present application, the reference prediction model is a K nearest neighbor model. The K nearest neighbor model is a model constructed based on the K-Nearest Neighbors (KNN) algorithm. The K nearest neighbor model classifies data based on distance and can effectively distinguish different driving styles. At the same time, compared with the deep learning model, the K nearest neighbor model has a faster output speed. Specifically, the KNN algorithm is used to construct a K nearest neighbor model cluster based on the driving data corresponding to the training set, and based on the accuracy of the output data of the K nearest neighbor model under different K values ​​under cross-validation conditions, the K value corresponding to the highest accuracy is selected as the K value of the reference prediction model. At the same time, the voting weights of the K neighbor samples can be set during the verification process, and the closer the distance, the higher the corresponding weight value. Among them, the accuracy of the reference prediction model is greater than the second preset value. In at least one embodiment of the present application, the second preset value is 85%.

[0166] The result detection unit 33 is electrically connected to the first prediction unit 31 and the second prediction unit 32. The result detection unit 33 is configured to determine and output a target driving style based on the core driving style and the reference prediction model.

[0167] The core prediction model may misjudge the distinction between robust and soft driving styles. To improve the accuracy of the prediction results, this application combines the core driving style and the reference driving style to determine the target driving style. This corrects the core driving style output by the deep learning model and improves the reliability of the driving style prediction results. After the calibration operation, the accuracy of the core prediction model is greater than a third preset value. The third preset value is greater than the first preset value and greater than the second preset value. In at least one embodiment of the present application, the third preset value is 95%.

[0168] The result detection unit 33 determines whether the core driving style is consistent with the reference driving style. When the core driving style is inconsistent with the reference driving style, the result detection unit 33 obtains a corresponding verification value based on the core driving style and the reference driving style.

[0169] In the first embodiment of the present application, when the core driving style is the first driving style and the reference driving style is the second driving style, the probability value of the second driving style is determined as the primary verification value, and the probability value of the third driving style is determined as the secondary verification value.

[0170] In a second embodiment of the present application, when the core driving style is the first driving style and the reference driving style is the third driving style, the probability value of the first driving style is determined as the primary verification value, the probability value of the third driving style is determined as the secondary verification value, and the probability value of the second driving style in the core prediction data is used as the final verification value.

[0171] In a third embodiment of the present application, when the core driving style is the second driving style and the reference driving style is the first driving style, the probability value of the first driving style is determined as the primary verification value, and the probability value of the third driving style is determined as the secondary verification value.

[0172] In a fourth embodiment of the present application, when the core driving style is the second driving style and the reference driving style is the third driving style, the probability value of the third driving style is determined as the primary verification value, the probability value of the second driving style is determined as the secondary verification value, and the probability value of the first driving style is determined as the final verification value.

[0173] In a fifth embodiment of the present application, when the core driving style is the third driving style and the reference driving style is the first driving style, the probability value of the third driving style is determined as the primary verification value, and the probability value of the second driving style is determined as the secondary verification value.

[0174] In a sixth embodiment of the present application, when the core driving style is the third driving style and the reference driving style is the second driving style, the probability value of the second driving style is determined as the primary verification value, and the probability value of the first driving style is determined as the secondary verification value.

[0175] The result detection unit 33 further determines the driving style based on the obtained verification value and outputs it.

[0176] In the first embodiment of the present application, when the core driving style is the first driving style and the reference driving style is the second driving style, the probability value of the second driving style serves as the primary check value, and the probability value of the third driving style serves as the secondary check value, the result detection unit 33 is configured to determine whether the primary check value is greater than or equal to a first primary threshold. If the primary check value is greater than or equal to the first primary threshold, the result detection unit 33 further determines whether the secondary check value is greater than or equal to a first secondary threshold. If the secondary check value is greater than or equal to the first secondary threshold, the result detection unit 33 sets the third driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style. If the secondary check value is less than the first secondary threshold, the result detection unit 33 sets the second driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style. If the primary check value is less than the first primary threshold, the result detection unit 33 further determines whether the secondary check value is greater than or equal to a second secondary threshold. If the secondary check value is greater than or equal to the second secondary threshold, the result detection unit 33 sets the third driving style as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style. When the secondary check value is less than the second secondary threshold, the result detection unit 33 further determines whether the secondary check value is greater than or equal to the third secondary threshold. If the secondary check value is greater than or equal to the third secondary threshold, the result detection unit 33 uses the second driving style as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style. When the secondary check value is less than the third secondary threshold, the result detection unit 33 uses the core driving style as the target driving style. That is, the core driving style output by the core prediction model is accurate and no correction operation is required. In at least one embodiment of the present application, the first primary threshold is 7%, the first secondary threshold is 25%, the second secondary threshold is 40%, and the third secondary threshold is 20%.

[0177] In the second embodiment, the core driving style is the first driving style and the reference driving style is the third driving style. The probability value of the first driving style serves as the primary check value, and the probability value of the third driving style serves as the secondary check value. The result detection unit 33 is configured to determine whether the primary check value is less than or equal to the second primary threshold. If the primary check value is less than or equal to the second primary threshold, the result detection unit 33 further determines whether the secondary check value is greater than or equal to the fourth secondary threshold. If the secondary check value is greater than or equal to the fourth secondary threshold, the result detection unit 33 uses the third driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the third driving style. If the secondary check value is less than the fourth secondary threshold, the result detection unit 33 uses the second driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the first driving style to the second driving style. If the primary check value is greater than the second primary threshold, the result detection unit 33 uses the core driving style as the target driving style. In other words, the core driving style output by the core prediction model is correct, and no correction operation is required. In at least one embodiment of the present application, the second main threshold is 95.3% and the fourth secondary threshold is 45%.

[0178] In the third embodiment, the core driving style is the second driving style and the reference driving style is the first driving style. The probability value of the first driving style serves as the primary check value, and the probability value of the third driving style serves as the secondary check value. The result detection unit 33 is configured to determine whether the primary check value is greater than or equal to a third primary threshold. If the primary check value is greater than or equal to the third primary threshold, the result detection unit 33 uses the first driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the first driving style. If the primary check value is less than the third primary threshold, the result detection unit 33 further determines whether the secondary check value is greater than or equal to a fifth secondary threshold. If the secondary check value is greater than or equal to the fifth secondary threshold, the result detection unit 33 uses the third driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style. If the secondary check value is less than the fifth secondary threshold, the result detection unit 33 uses the core driving style as the target driving style. This indicates that the core driving style output by the core prediction model is correct, and no correction operation is required. In at least one embodiment of the present application, the third primary threshold is 7%, the fifth secondary threshold is 40%, and the fifth secondary threshold is 21.5%.

[0179] In the fourth embodiment, the core driving style is the second driving style and the reference driving style is the third driving style. The probability value of the third driving style serves as the primary check value, the probability value of the second driving style serves as the secondary check value, and the probability value of the first driving style serves as the final check value. The result detection unit 33 is configured to determine whether the primary check value is greater than or equal to a fourth primary threshold. When the primary check value is greater than or equal to the fourth primary threshold, the result detection unit 33 determines whether the secondary check value is greater than or equal to a sixth secondary threshold. When the secondary check value is greater than or equal to the sixth secondary threshold, the result detection unit 33 uses the first driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the first driving style. When the secondary check value is less than the sixth secondary threshold, the result detection unit 33 uses the third driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style. When the primary check value is less than the fourth primary threshold, the result detection unit 33 determines whether the final check value is greater than or equal to the first final threshold. When the final-level check value is less than the first final-level threshold, the result detection unit 33 uses the third driving style as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the second driving style to the third driving style. When the final-level check value is greater than or equal to the first final-level threshold, the result detection unit 33 uses the core driving style as the target driving style. That is, the core driving style output by the core prediction model is correct, and no correction operation is required. In at least one embodiment of the present application, the fourth primary threshold is 5%, the sixth secondary threshold is 29.8%, and the first final-level threshold is 77%.

[0180] In the fifth embodiment, the core driving style is the third driving style and the reference driving style is the first driving style. The probability value of the third driving style serves as the primary check value, and the probability value of the second driving style serves as the secondary check value. The result detection unit 33 is configured to determine whether the primary check value is greater than or equal to the fifth primary threshold. If the primary check value is greater than or equal to the fifth primary threshold, the result detection unit 33 determines whether the secondary check value is greater than or equal to the seventh secondary threshold. If the secondary check value is greater than or equal to the seventh secondary threshold, the result detection unit 33 uses the second driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the second driving style. If the secondary check value is less than the seventh secondary threshold, the result detection unit 33 uses the first driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the first driving style. If the primary check value is less than the fifth primary threshold, the result detection unit 33 uses the core driving style as the target driving style. This indicates that the core driving style output by the core prediction model is correct, and no correction operation is required. In at least one embodiment of the present application, the fifth primary threshold is 55% and the seventh secondary threshold is 16%.

[0181] In the sixth embodiment, the core driving style is the third driving style and the reference driving style is the second driving style. The probability value of the second driving style serves as the primary check value, and the probability value of the first driving style serves as the secondary check value. The result detection unit 33 is configured to determine whether the primary check value is greater than or equal to the sixth primary threshold. When the primary check value is greater than or equal to the sixth primary threshold, the result detection unit 33 determines whether the secondary check value is greater than or equal to the eighth secondary threshold. When the secondary check value is greater than or equal to the eighth secondary threshold, the result detection unit 33 sets the first driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the first driving style. When the secondary check value is less than the eighth secondary threshold, the result detection unit 33 sets the second driving style as the target driving style. In other words, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the second driving style. When the primary check value is less than the sixth primary threshold, the result detection unit 33 determines whether the secondary check value is greater than or equal to the ninth secondary threshold. When the secondary check value is greater than or equal to the ninth secondary threshold, the result detection unit 33 sets the second driving style as the target driving style. That is, a correction operation is performed to correct the core driving style output by the core prediction model from the third driving style to the second driving style. When the secondary check value is less than the ninth secondary threshold, the result detection unit 33 uses the core driving style as the target driving style. This means that the core driving style output by the core prediction model is correct, and no correction operation is required. In at least one embodiment of the present application, the sixth primary threshold is 21%, the eighth secondary threshold is 12.5%, and the ninth secondary threshold is 5%.

[0182] Verification module 40 is configured to determine whether the subjectively evaluated driving style is consistent with the target driving style. In at least one embodiment of the present application, the subjectively evaluated driving style includes a driving style provided by the user based on subjective judgment via the input / output device 4 of vehicle 100. The evaluation data may be user-provided evaluation data when vehicle 100 passes through key locations (such as turns, roundabouts, and gates) or travels on specific experimental sections (corresponding to different road types). If the subjectively evaluated driving style is inconsistent with the target driving style, driving data corresponding to the subjectively evaluated style is obtained to update the training, validation, and test sets.

[0183] Based on the aforementioned driving style prediction system 1, a core prediction model and a reference prediction model are used to analyze driving data from two dimensions. A target driving style is determined based on the core and reference driving styles. This reduces the probability of misjudgments by the core prediction model during driving style prediction and improves the reliability of driving style prediction results. Furthermore, the combination of the core and reference prediction models to optimize prediction results reduces the complexity of building the core prediction model compared to optimizing the core prediction model.

[0184] See also Figure 10 , which is a block diagram of a vehicle 100 according to at least one embodiment of the present application. Vehicle 100 is a hybrid vehicle. It is understood that the present application does not limit the type of vehicle 100, and it may be, for example, a plug-in hybrid vehicle or a gasoline-electric hybrid vehicle.

[0185] Vehicle 100 includes a driving style prediction system 1, a sensor 2, a vehicle controller 3, an input / output device 4, a processor 5, a memory 6, and a communication interface 7. The driving style prediction system 1, the sensor 2, the vehicle controller 3, the input / output device 4, the processor 5, the memory 6, and the communication interface 7 can be connected via a communication bus and communicate with each other.

[0186] The specific contents of the driving style prediction system 1, the sensor 2, the vehicle controller 3 and the input / output device 4 can be found in the specific description of the above-mentioned driving style prediction method, and will not be described in detail here.

[0187] The processor 5 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the above program.

[0188] The memory 6 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 6 can exist independently and be connected to the processor 5 via a bus. The memory 6 can also be integrated with the processor 5.

[0189] The memory 6 is used to store program instructions for executing the above scheme, and the execution is controlled by the processor 5. The processor 5 is used to execute the program instructions stored in the memory 6. The program instructions stored in the memory 6 can be executed. Figure 1 as well as Figure 2 Part or all of the steps of the driving style prediction method described in.

[0190] The communication interface 7 is used to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Networks (WLAN), etc.

[0191] In at least one embodiment of the present application, a computer-readable storage medium (not shown) is further provided, in which computer-readable instructions are stored. The computer-readable instructions are executed by a processor in an electronic device to implement the driving style prediction method of any of the above embodiments.

[0192] In the several embodiments provided herein, it should be understood that the disclosed systems and methods can be implemented in other ways. Modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, i.e., they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the objectives of the present embodiments based on actual needs.

[0193] In addition, the functional modules in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0194] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices listed in the specification may also be implemented by a single unit or device through software or hardware. Terms such as first and second are used to indicate names and do not imply any particular order.

[0195] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, as long as they are within the scope of the essence of the present application, appropriate changes and modifications made to the above embodiments should fall within the scope of protection claimed in the present application.

Claims

1. A driving style prediction method, characterized by: The driving style prediction method includes: Acquire driving data of the vehicle during driving; Preprocessing the driving data to obtain a training set, a validation set, and a test set, wherein the sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model; Inputting the sample data in the test set into the core prediction model to obtain a core driving style; Inputting the sample data in the test set into the reference prediction model to obtain a reference driving style; A target driving style is determined and outputted according to the core driving style and the reference driving style.

2. The driving style prediction method according to claim 1, wherein: The driving style prediction method further includes: When it is determined that the target driving style is inconsistent with the subjective evaluation style input by the user, obtaining driving data corresponding to the subjective evaluation style; The driving data corresponding to the subjective evaluation style is preprocessed to update the training set, the validation set, and the test set.

3. The driving style prediction method according to claim 1, wherein: The determining and outputting a target driving style according to the core driving style and the reference driving style includes: When the core driving style is inconsistent with the reference driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style; Determine the target driving style according to the obtained verification value and output it; When the core driving style is consistent with the reference driving style, the core driving style or the reference driving style is determined as the target driving style and output.

4. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is a first driving style and the reference driving style is a second driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the second driving style as the primary verification value, and determining the probability value of the third driving style as the secondary verification value; The determining and outputting the target driving style according to the obtained verification value includes: When the primary check value is greater than or equal to a first primary threshold and the secondary check value is greater than or equal to a first secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is greater than or equal to the first primary threshold and the secondary check value is less than the first secondary threshold, determining the second driving style as the target driving style and outputting the result; When the primary check value is less than the first primary threshold and the secondary check value is greater than or equal to a second secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is less than the first primary threshold, the secondary check value is less than the second secondary threshold, and the secondary check value is greater than or equal to a third secondary threshold, determining the second driving style as the target driving style and outputting the result; When the primary check value is less than the first primary threshold and the secondary check value is less than the third secondary threshold, determining the core driving style as the target driving style and outputting the target driving style; The second secondary threshold is greater than the first secondary threshold, and the third secondary threshold is less than the second secondary threshold.

5. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the first driving style and the reference driving style is the third driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the first driving style as the primary verification value, and determining the probability value of the third driving style as the secondary verification value; The determining and outputting the target driving style according to the obtained verification value includes: When the primary check value is less than or equal to a second primary threshold and the secondary check value is greater than or equal to a fourth secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is less than or equal to the second primary threshold and the secondary check value is less than the fourth secondary threshold, determining the second driving style as the target driving style and outputting the target driving style; When the main verification value is greater than the second main threshold, the core driving style is determined as the target driving style and output.

6. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the second driving style and the reference driving style is the first driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the first driving style as the primary verification value, and determining the probability value of the third driving style as the secondary verification value; The determining and outputting the target driving style according to the obtained verification value includes: When the main verification value is greater than or equal to a third main threshold, determining the first driving style as the target driving style and outputting the target driving style; When the primary check value is less than the third primary threshold and the secondary check value is greater than or equal to a fifth secondary threshold, determining the third driving style as the target driving style and outputting the result; When the primary check value is smaller than the third primary threshold and the secondary check value is smaller than the fifth secondary threshold, the core driving style is determined as the target driving style and output.

7. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value, a secondary check value, and a final check value; When the core driving style is the second driving style and the reference driving style is the third driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the third driving style as the primary verification value, determining the probability value of the first driving style as the secondary verification value, and determining the probability value of the second driving style as the final verification value; The determining and outputting the target driving style according to the obtained verification value includes: When the primary check value is greater than or equal to a fourth primary threshold and the secondary check value is greater than or equal to a sixth secondary threshold, determining the first driving style as the target driving style and outputting the target driving style; When the primary check value is less than the fourth primary threshold and the secondary check value is less than the sixth secondary threshold, determining the third driving style as the target driving style and outputting the target driving style; When the main check value is less than the fourth main threshold and the final check value is less than the first final threshold, determining the third driving style as the target driving style and outputting the target driving style; and When the main verification value is smaller than the fourth main threshold and the final-stage verification value is greater than or equal to the first final-stage threshold, the core driving style is determined as the target driving style and output.

8. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the third driving style and the reference driving style is the first driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the third driving style as the primary verification value, and determining the probability value of the second driving style as the secondary verification value; The determining and outputting the target driving style according to the obtained verification value includes: When the primary check value is less than or equal to a fifth primary threshold and the secondary check value is greater than or equal to a seventh secondary threshold, determining the second driving style as the target driving style and outputting the target driving style; When the primary check value is less than or equal to the fifth primary threshold and the secondary check value is less than the seventh secondary threshold, determining the first driving style as the target driving style and outputting the target driving style; and When the main verification value is greater than the fifth main threshold, the core driving style is determined as the target driving style and output.

9. The driving style prediction method according to claim 3, wherein: The core driving style and the reference driving style are respectively one of a first driving style, a second driving style, and a third driving style, and the check value includes a primary check value and a secondary check value; When the core driving style is the third driving style and the reference driving style is the second driving style, obtaining a corresponding verification value according to the core driving style and the reference driving style includes: determining the probability value of the second driving style as the primary verification value, and determining the probability value of the first driving style as the secondary verification value; The determining and outputting the target driving style according to the obtained verification value includes: determining the first driving style as the target driving style when the primary check value is greater than or equal to a sixth primary threshold and the secondary check value is greater than or equal to an eighth secondary threshold; determining the second driving style as the target driving style when the primary check value is greater than or equal to the sixth primary threshold and the secondary check value is less than the eighth secondary threshold; and When the primary check value is less than the sixth primary threshold and the secondary check value is greater than or equal to a ninth secondary threshold, determining the second driving style as the target driving style; and When the primary check value is less than the sixth primary threshold and the secondary threshold is less than or equal to the ninth secondary threshold, determining the core driving style as the target driving style; The ninth secondary threshold is smaller than the eighth secondary threshold.

10. A driving style prediction system, characterized in that: The driving style prediction system includes: A data acquisition module is used to obtain driving data of the vehicle during driving; A data processing module, configured to pre-process the driving data to obtain a training set, a validation set, and a test set; wherein the sample data in the training set and the validation set are used to train the core prediction model and the reference prediction model; Prediction module, used to: Inputting the sample data in the test set into the core prediction model to obtain a core driving style; Inputting the sample data in the test set into the reference prediction model to obtain a reference driving style; A target driving style is determined and outputted according to the core driving style and the reference driving style.

11. A vehicle, characterized in that: The vehicle includes a processor and a memory, the memory is used to store a plurality of program instructions, and when the processor calls the program instructions, the driving style prediction method according to any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions that can be executed by at least one processor, and when the computer-executable instructions are executed by the at least one processor, the driving style prediction method according to any one of claims 1 to 9 is implemented.

Images