CN105426638A - Driver behavior characteristic identification device - Google Patents

Abstract

The invention provides a driver behavior characteristic identification device, which comprises a real vehicle driving simulator experiment platform and a driver behavior characteristic identification method. Wherein: the real vehicle driving simulator experimental platform includes a real vehicle driving simulator experimental platform main body, a console, a vehicle dynamics simulation model, a real-time simulation system, a steering force simulation system and a sensor system; the driver behavior characteristic identification The method is to first complete the establishment and verification of the driver behavior characteristic identification model, and then use the established driver behavior characteristic identification model to identify the driver behavior characteristic. The real vehicle driving simulator experiment platform provides the driver with a test platform close to the real vehicle to obtain data samples and prepare for the establishment of the driver behavior characteristic identification model; it fully considers the interaction interface between the driver and the car, through The simulation of vision, hearing and touch gives the driver the feeling of driving a real car, and the effect is realistic.

Description

A driver behavior characteristic recognition device

technical field

The invention belongs to the field of automobiles, in particular to a device for identifying driver behavior characteristics.

Background technique

Classifying and identifying driver behavior characteristics is a necessary prerequisite for the development of personalized driver assistance systems and the realization of advanced technologies such as "cars adapt to people" ideal car dynamics. When different types of drivers drive a vehicle, the ideal response characteristics of the vehicle are different. For example, racing drivers like fast acceleration dynamics, and novices like steering performance with constant steering characteristics. The driving assistance system (vehicle control system) first recognizes the behavior characteristics of the driver, and then changes the response characteristics of the vehicle by changing the relevant control parameters or switching the control strategy, so as to realize the active adaptation of the vehicle to the specific type of behavior characteristics of the driver.

In the past, the identification of driver behavior characteristics was based on actual vehicles. The driver drives these actual vehicles, and the driver's behavior characteristics are identified according to the driver's driving behavior in a given driving environment and the state of the vehicle. On the one hand, this driving process will consume fuel and a large number of motion sensors, which is economical; on the other hand, it may cause problems if drivers with unknown driving proficiency, uneven driving levels, and different driving styles participate in the actual driving. Dangerous, safety is not guaranteed; Finally, due to the impact of the environment, the reproducibility of the test is also poor.

Chinese invention patent application 201080059425.X, the driver drives a straddle vehicle to complete the turning movement, and the driver's characteristics are judged according to the state quantity of the vehicle and the driver's head movement. The disadvantage is that the straddle vehicle required for the test is a special two-wheeled vehicle that needs to be customized; the test process requires a large number of expensive sensors to obtain vehicle, road, and road surface information; and the calculation process is complicated and costly.

Chinese invention patent application 201110101625.3 relates to a method for adapting the driving characteristics of a vehicle to driver changes. It stores driving parameters representing specific types of driving styles, and also stores safety algorithms corresponding to drivers of various types of driving styles. The idea is that the same car, by identifying the driver's driving style, activates (or newly generates) its corresponding safety algorithm to achieve the purpose of "car adapting to people". The limitation of this invention is that it only relies on the state quantity of the vehicle to describe the driving style, ignoring the driver's operation behavior; in fact, the driver's operation behavior also includes a lot of content that reflects the driver's driving style. Although there is information redundancy between it and the vehicle state quantity, it should not be discarded easily; this invention only proposes a logical process of "car adapting to people", and does not clearly give the specific process of driver style identification and methods.

Contents of the invention

The present invention aims to provide a device for identifying driver behavior characteristics. Through the construction of a real vehicle driving simulator test platform and the development of a driver behavior identification method, low-cost, safe and efficient identification of driver behavior characteristics can be realized.

In order to achieve the above object, the technical solution adopted by the present invention is that the hardware part of the driver behavior characteristic identification device is a real vehicle driving simulator test platform built based on CarSimRT/Simulink/dSPACE, including the real vehicle driving simulator test platform main body, control Key components such as platform, vehicle dynamics simulation model, real-time simulation system, steering force simulation system and sensor system. A number of drivers are selected to conduct a large number of tests under different working conditions on the real vehicle driving simulator test platform to obtain data samples to fully mine the driver's behavior characteristics and prepare for the establishment of a driver behavior characteristic identification model.

Among them, the main body of the real vehicle driving simulator test platform is a modified domestic car, which is used to provide the driver with a realistic driving environment and is the basis for the installation of related instruments and equipment; the console includes a main control computer, and The ring screen is composed of 3 monitors. The main control computer of the console mainly conducts the test, development and verification of the vehicle dynamics model and related control systems. The ring screen composed of 3 monitors displays the dynamic traffic scene provided by CarSim in real time, which is convenient to be next to the console. The experimenter observes the test situation; Utilize CarSimRT/Simulink/dSPACE to set up the real-time vehicle dynamics simulation system model; The real vehicle driving simulator platform that the present invention builds adopts one of the products of real-time simulation system dSPACE DS1006PPC processor board, DS1006 processor board Perform high-speed communication with all dSPACE I/O boards through the PHS bus; adopt a force-sensing simulation system based on the C-EPS structure, and the steering force-sensing module receives the steering kingpin torque and vehicle speed calculated by the CarSim vehicle dynamics model, And the steering wheel angle and force sense motor torque collected by the sensor system, calculate the target current command of the steering force sense analog motor, and accurately control the torque of the motor through the motor vector control strategy, and finally provide the driver with road feeling when turning ; The sensor system mainly includes steering wheel angle sensor, accelerator pedal position sensor and brake pedal position sensor.

In order to achieve the above purpose, the present invention develops a driver behavior characteristic identification method. Its main process includes: based on the real vehicle driving simulator test platform, design the steering, braking, and acceleration test conditions; select test personnel to conduct the test and collect data; analyze the behavior of steering, braking, and acceleration, and select representative driving conditions. The parameters of each driver's handling behavior characteristics, that is, the characteristic parameters of steering behavior - steering wheel speed, standard deviation of steering wheel angle and average vehicle speed, and the characteristic parameters of braking behavior - maximum brake pedal opening, brake pedal speed and collision avoidance time TTC, characteristic parameters of acceleration behavior - maximum accelerator pedal opening, accelerator pedal speed and driving speed; use MATLAB to write programs to extract the above characteristic parameters from the collected test data; cluster the characteristic parameters based on the K-means algorithm, and then Divide the driver's steering, braking, and acceleration behavior characteristics into three categories: cautious, general, and aggressive, and provide data samples for building a driver behavior characteristic identification model; use BP neural network, combined with the above clustering results to establish a driver Behavioral characteristic identification model; finally complete the verification of the accuracy and predictive ability of the established driver behavior characteristic identification model. After that, for any driver, according to his operation behavior on the real vehicle driving simulator test platform and the motion state of the virtual vehicle, multiple sets of characteristic parameters can be extracted and sent to the established driver behavior characteristics Identify the model to obtain the identification results of each set of characteristic parameters, count the frequency and proportion of cautious, general, and aggressive driving behaviors in the identification results, and identify the driver's behavior characteristics according to the principle of maximum membership .

The K-means algorithm is a clustering method based on unsupervised learning between data dissimilarities, and the specific steps of the clustering process are as follows:

(1) Input the number K of clusters and the data sample D to be classified;

(2) Randomly select K elements or the first K elements from D as the initial cluster center;

(3) Utilize the Euclidean distance to calculate the distances between the remaining elements and the K cluster centers, and divide these elements according to the principle of the minimum distance;

(4) Take the arithmetic mean of all elements in the K clusters as the new cluster center;

(5) Re-cluster all the elements in the data sample D according to the new cluster center;

(6) Determine whether there is element exchange in each cluster, if yes, repeat (5), if not, end;

(7) Output K clusters.

The BP neural network is a multi-layer feed-forward network trained according to the error back propagation algorithm, which can learn and store a large number of input-output pattern mapping relationships without revealing the mathematical equations describing the mapping relationships in advance. The input of the BP neural network is a characteristic parameter, and the corresponding output is the behavior characteristic type represented by the characteristic parameter. The BP neural network is used to establish the identification model of driver's behavior characteristics.

Description of drawings:

Fig. 1 is the frame diagram of the real vehicle driving simulator test platform system of the present invention.

Fig. 2 is an overall structure diagram of the force-sensing simulation system of the present invention.

Fig. 3 is a process diagram of establishing a driver behavior characteristic identification model in the present invention.

Fig. 4 is a flowchart of the K-means algorithm of the present invention.

Fig. 5 is a graph showing the classification results of steering behavior characteristics of the present invention.

Fig. 6 is a schematic diagram of the BP neural network structure of the steering behavior characteristic identification model of the present invention.

Fig. 7 is a diagram of the accuracy verification process of the driver behavior characteristic identification model of the present invention.

Fig. 8 is a process diagram of the driver's behavior characteristic prediction ability verification process in the present invention.

Fig. 9 is a process diagram of driver behavior characteristic identification in the present invention.

Fig. 10 is a diagram of the identification results of the cautious driver's behavior characteristics according to the present invention.

Fig. 11 is a diagram of identification results of general driver behavior characteristics of the present invention.

Fig. 12 is a diagram of identification results of aggressive driver behavior characteristics of the present invention.

Fig. 13 is a diagram of the identification result of driver No. 1's behavior characteristic in the present invention.

Fig. 14 is a diagram of the recognition results of driver No. 2 behavior characteristics of the present invention.

Fig. 15 is a diagram of the identification results of driver No. 3 behavior characteristics of the present invention.

detailed description:

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

Fig. 1 represents the system framework of the real vehicle driving simulator test platform of the present invention, comprising: real vehicle driving simulator test platform main body (1), console (2), vehicle dynamics simulation model (3), real-time simulation system ( 4), steering force simulation system (5) and sensor system (6) and other key components.

Among them, the real vehicle driving simulator main body (1) provides a realistic driving environment for the driver, and is an installation platform for various auxiliary instruments and equipment; there is a main control computer on the console (2), and the main control computer mainly uses The real-time dynamics simulation software CarSimRT, the algorithm development software MATLAB/Simulink and the real-time simulation system dSPACE, etc. are used to test, develop and verify the vehicle dynamics model and related control systems. The ring-screen real-time display composed of three monitors on the console is provided by CarSim. The dynamic traffic scene is convenient for the test personnel next to the console to observe the test situation; the vehicle type selection and parameter setting, simulation working condition design and simulation parameter setting are carried out in CarSimRT, and the vehicle dynamics simulation model is established in conjunction with MATLAB/Simulink/RTI (3 ), compile and generate a real-time simulation program and load it into the dSPACE measurement and control platform; use CarSimRT/Simulink/dSPACE to establish a real-time vehicle dynamics simulation system model (4); the steering force simulation system (5) receives the CarSim vehicle dynamics model calculation The steering kingpin torque and vehicle speed obtained by the sensor system, as well as the steering wheel angle and force-sensing motor torque collected by the sensor system, are calculated to obtain the target current command of the steering force-sensing analog motor, and the motor is accurately controlled by the motor vector control strategy. Ultimately provide road sense for the driver when turning; the sensor system (6) collects the driver's steering manipulation signal, operates the accelerator pedal behavior information, manipulates the brake pedal behavior information, and provides the driver's various manipulations for the CarSim dynamics simulation model Signal.

Fig. 3 is a process diagram of establishing a driver behavior characteristic identification model in the present invention. The present invention divides the driver's behavior characteristics into steering characteristics, braking characteristics and acceleration characteristics, selects a number of drivers to conduct steering, braking and acceleration tests on the real vehicle driving simulator test platform, and collects the driver's manipulation behavior signals and The vehicle's motion state information, by analyzing the driver's behavior, extracts the characteristic parameters that can characterize the driver's steering, braking, and acceleration behavior characteristics, and clusters the characteristic parameters, so as to realize the classification of driver behavior characteristics and obtain the characteristics of each control. The characteristic parameter samples included in different characteristic types of behaviors, and finally establish the characteristic identification model of each manipulation behavior based on BP neural network.

Fig. 7 is the process of verifying the accuracy of the driver behavior characteristic identification model of the present invention. Select a number of drivers with known behavioral characteristics to conduct steering, braking, and acceleration tests on the real vehicle driving simulator test platform, collect the driver's manipulation behavior signals and the vehicle's motion state information, and extract , The characteristic parameters of the acceleration behavior characteristics are sent to the characteristic identification models of each maneuvering behavior, and the identification results of the driver's behavior characteristics are obtained to verify the reliability of the previously established driver behavior characteristic identification model.

Fig. 8 is the process of verifying the predictive ability of the identification model of the driver's behavior characteristic in the present invention. Select a number of drivers whose behavior characteristics are unknown to conduct steering, braking, and acceleration tests on the real vehicle driving simulator test platform, collect the driver's manipulation behavior signals and vehicle motion state information, and extract drivers that can represent the driver's steering, braking, and acceleration tests. The characteristic parameters of the acceleration behavior characteristics are sent to the characteristic identification model of each maneuvering behavior to obtain the identification result of the driver's behavior characteristics.

Fig. 9 is the identification process of the driver's behavior characteristics in the present invention. On the premise that the establishment of the driver behavior characteristic model is completed, and the accuracy verification of the model meets the requirements and the prediction ability meets the expectations, the model can be used to identify the behavior characteristics of any driver according to this process.

Specific embodiment verification:

In order to verify the feasibility of the driver behavior characteristic identification device provided by the present invention, this embodiment demonstrates the complete process of realizing driver behavior characteristic identification by using a real vehicle driving simulator test platform, combined with a driver behavior characteristic identification method. Since the idea and process of identification of steering behavior characteristics, braking behavior characteristics and acceleration behavior characteristics are basically similar, and based on the same real vehicle driving simulator test platform, only the set working conditions and selected characteristic parameters are different. Therefore, the present invention only takes the driver's steering behavior characteristic recognition process as a typical case to describe the driver behavior characteristic recognition method.

Select 13 drivers to test on the real vehicle driving simulator test platform, extract characteristic parameters from the collected test data - steering wheel speed, steering wheel angle standard deviation and average vehicle speed, and use K-means for clustering , to obtain data samples of drivers' steering behaviors classified as cautious, general and aggressive. A total of 543 steering feature parameters were extracted in this example, and three data sets were obtained after clustering, as shown in Figure 5. The number of elements in each data set is 210, 198, and 135, respectively, representing the driver's cautious, general, and aggressive steering characteristics, and the three cluster centers are [0.10420.14090.3228], [0.14490.18130.6110 ], [0.32550.42300.6305], after denormalization, the cluster centers are [13.646711.166428.4855], [15.953611.570835.4012], [46.285328.380647.4968]. A driver with cautious steering characteristics will use a lower speed and a smaller steering wheel speed to corner, and the fluctuation of the steering wheel angle will also be small; a driver with aggressive steering characteristics will generally have a higher speed when cornering, but the steering wheel speed will also be biased. High, and will always adjust the steering wheel, so there will be a higher standard deviation of the steering wheel angle; while the steering wheel speed, standard deviation of the steering wheel angle and vehicle speed of the driver with general steering characteristics are all in the cautious type and aggressive type when cornering between.

The driver's steering behavior characteristic type is determined by counting the proportion of the driver's steering behavior characteristic parameter samples distributed in the three types. The statistical results of the 13 testers are shown in Table 1. It can be seen from the ratio that each tester has more than 50% of the samples belonging to one of the three types of driver characteristics, so the type of driver can be determined, as shown in Table 1, 5 cautious type, general Type 6, radical type 2. The obtained data samples of the three kinds of driver's steering behavior characteristics provide a basis for establishing a driver's steering behavior characteristic identification model.

Table 1 Types of steering behavior characteristics of test personnel

The present invention determines and selects the BP neural network as the modeling method of the identification model of the driver's behavior characteristics. Using various types of data samples obtained in driver behavior classification, a driver behavior characteristic identification model is established based on BP neural network. The data samples of the three types of driver behavior characteristics obtained before are used as the input of the BP neural network, and the output is encoded using the "1 out of n" representation, that is, the cautious, general and aggressive drivers The output quantities of are taken as 100, 010 and 001 respectively. Since the number of neurons in the input layer and output layer of the BP neural network is determined by the dimensions of the input and output respectively, the BP neural network is a network with 3 inputs and 3 outputs. In order to enhance the mapping capability of the network and improve the training accuracy of the network, the present invention repeatedly adjusts the number of nodes in the two hidden layers, and finally determines the number of nodes to be 5 and 3 respectively. After the above design, the obtained network structure diagram is shown in Figure 6. Among them, P1, P2, and P3 are the three dimensions of the feature parameter samples, iw1, iw2, and iw3 are the weights from P1, P2, and P3 to the first hidden layer, respectively, and lw1, lw2 are the weights from the first hidden layer to the second hidden layer, respectively. layer, the weight of the second hidden layer to the output layer, b1, b2 and b3 are the thresholds of the first hidden layer, the second hidden layer and the output layer respectively, n1, n2 and n3 are the input layer, the first hidden layer and the The weighted calculation amount of the second hidden layer, f1, f2 and f3 are the S(sigmoid) transfer function respectively, a1, a2 and a3 are the output values of the first hidden layer, the second hidden layer and the output layer respectively, and a3 is a three-dimensional column vector, representing the type of behavior characteristics of the driver.

Accuracy verification process of identification model of driver's steering behavior characteristics. Three drivers whose steering behavior characteristics are known in Table 1 (where driver 1 is a cautious type, driver 2 is an ordinary type, and driver 3 is an aggressive type) are selected to perform steering on the real vehicle driving simulator test platform. Test, collect test data and extract steering characteristic parameters, use the established steering behavior characteristic identification model for identification, the results are shown in Table 2. It can be seen from the table that the steering behavior characteristics identification model can correctly identify the driver's steering behavior characteristics, and the identification results are consistent with the types of the three known test drivers, which shows that the steering behavior characteristics identification model has high accuracy. In order to illustrate the verification results more intuitively, the driver’s steering behavior operation data can be compared with the identification results, as shown in Figure 10, Figure 11, and Figure 12, which are respectively driver 1 (cautious type), driver 2 (general type), and driver 2 (general type). Identification results for driver 3 (aggressive).

Table 2 Accuracy verification results of driver steering behavior characteristics identification model

driver serial number cautious general type Radical Statistics 1 (cautious) 40 5 2 Cautious (85.1%) 2 (general type) 8 26 9 General type (60.5%) 3 (aggressive) 0 11 18 Aggressive (62.1%)

Verification of predictive ability of identification model of driver's steering behavior characteristics. Three drivers whose steering characteristics are unknown are selected, and the steering test is carried out on the real vehicle driving simulator test platform, the test data is collected and the steering characteristic parameters are extracted, and the steering characteristics of the driver are identified by the established steering behavior characteristic identification model. The results are shown in Table 3. It can be seen from the table that the steering behavior identification model identified more than 55% of the steering behaviors of the three drivers belonging to one of the cautious, general, and aggressive types, indicating that the steering behavior identification model has a good prediction ability. ability. In order to explain the prediction results more intuitively, the driver’s steering behavior manipulation data can be compared with the identification results. Figure 13, Figure 14, and Figure 15 respectively show the identification results of driver 1, driver 2, and driver 3

Table 3 The verification results of the predictive ability of the driver's steering behavior characteristics identification model

driver serial number cautious general type Radical forecast result 1 0 40 3 General type (93.0%) 2 56 1 1 Cautious (96.6%) 3 1 45 twenty three General type (65.2%)

Claims (7)

1. A driver's behavior characteristic identification device, is characterized in that, comprises following two parts: (1) real vehicle driving simulator test platform, the present invention builds real vehicle driving simulator test platform, carries out a large amount of tests under different working conditions, obtains data sample, to fully excavate the driver's behavioral characteristics, in order to set up the driver's behavioral characteristics The identification model is prepared; the real vehicle driving simulator test platform includes: real vehicle driving simulator test platform main body, console, vehicle dynamics simulation model, real-time simulation system, steering force simulation system and sensor system and other key components ; (2) driver's behavior characteristic identification method, described driver's behavior characteristic identification method comprises: set up driver's behavior characteristic identification model based on BP neural network, and the precision and predictive ability of driver's behavior characteristic identification model are verified; Afterwards, For any driver, the established driver behavior characteristic identification model can be used to identify the driver based on the operation behavior of the relevant test on the real vehicle driving simulator test platform and the motion state displayed by the virtual vehicle. behavioral characteristics are identified. 2. real vehicle driving simulator test platform according to claim 1, is characterized in that, mainly comprises: (1) real vehicle driving simulator test platform main body, described real vehicle driving simulator test platform main body is refitted A domestic car is used to provide the driver with a realistic driving environment, and is an installation platform for various auxiliary instruments and equipment; (2) console, which includes a main control computer and three monitors The main control computer mainly uses real-time dynamics simulation software CarSimRT, algorithm development software MATLAB/Simulink and real-time simulation system dSPACE to test, develop and verify vehicle dynamics models and related control systems; it consists of three displays The ring screen displays the dynamic traffic scene provided by CarSim in real time, which is convenient for the test personnel next to the console to observe the test situation; (3) the vehicle dynamics simulation model, which includes vehicle type selection and parameter setting in CarSimRT , simulation working condition design and simulation parameter setting, joint MATLAB/Simulink/RTI to establish vehicle dynamics control model, generate real-time simulation program after compilation and load in the dSPACE measurement and control platform; (4) real-time simulation system, the real-time simulation system Use CarSimRT/Simulink/dSPACE to establish a real-time vehicle dynamics simulation system model. The RTI interface of dSPACE receives the signals collected by the sensor system (including steering wheel signals, accelerator pedal signals, brake pedal signals and gear signals), and the SignalProcess module The signal is processed, and the steering wheel angle, the opening of the accelerator pedal, the gear position (R/N/D), the brake master cylinder pressure, the braking force of the handbrake to the two rear wheels and the braking force of the P gear to the two front wheels Input them into CarSim in turn, and CarSim outputs longitudinal vehicle speed, engine speed, left and right kingpin torque, longitudinal acceleration, lateral acceleration and other signals, among which, the longitudinal speed and engine speed are transmitted to the instrument panel through the RTI interface, and the left and right kingpin torque is used for calculation The target current provided to the steering force simulation system; (5) the steering force simulation system, which receives the steering kingpin torque and vehicle speed calculated by the CarSim vehicle dynamics model, and the steering force collected by the sensor system The disc rotation angle and force-sensing motor torque are calculated to obtain the target current command of the steering force-sensing analog motor, and the motor is accurately torque-controlled through the motor vector control strategy, finally providing road feeling for the driver when steering; (6) sensor system, The sensor system collects the driver's steering operation signal, the behavior information of the accelerator pedal operation and the behavior information of the brake pedal operation, and provides the driver's various manipulation signals for the CarSim dynamics simulation model. 3. according to claim 1 based on BP neural network to set up the driver's behavior characteristic identification model, it is characterized in that: the driver's behavior characteristics are divided into steering characteristics, braking characteristics and acceleration characteristics, select some drivers in the actual vehicle Carry out steering, braking, and acceleration tests on the driving simulator test platform, collect the driver's manipulation behavior signals and the vehicle's motion state, and analyze the driver's behavior to extract features that can characterize the driver's steering, braking, and acceleration behavior characteristics Parameters, and the characteristic parameters are clustered, so as to realize the classification of driver behavior characteristics, obtain the characteristic parameter samples included in the different characteristic types of each maneuvering behavior, and finally establish the characteristic identification model of each maneuvering behavior based on the BP neural network. 4. The characteristic parameter according to claim 3 is clustered, it is characterized in that the determined steering behavior characteristic parameters are: steering wheel speed, steering wheel angle standard deviation and average vehicle speed, and the braking behavior characteristic parameters are: maximum Brake pedal opening, brake pedal speed, and collision avoidance time TTC, and the characteristic parameters of acceleration behavior are: maximum accelerator pedal opening, accelerator pedal speed, and driving speed; the clustering method used is the k-means clustering algorithm, and the prior The input D data objects are divided into K clusters so that the obtained clusters satisfy: the similarity of objects in the same cluster is high; while the similarity of objects in different clusters is small. 5. BP neural network according to claim 3 is characterized in that, is a kind of multi-layer feed-forward network trained by error backpropagation algorithm, can learn and store a large amount of input-output pattern mapping relations, without prior knowledge The mathematical equation describing this mapping relationship is revealed; the input of the BP neural network is a characteristic parameter, and the corresponding output is the behavior characteristic type corresponding to the characteristic parameter. 6. The driver's behavior characteristic identification model accuracy verification according to claim 1 is characterized in that some drivers whose behavior characteristics are known are selected to carry out steering, braking and acceleration tests on the real vehicle driving simulator test platform, and collect driving The driver's maneuvering behavior signal and the vehicle's motion state are extracted from it to characterize the driver's steering, braking, and acceleration behavioral characteristics, and then sent to the established characteristic identification model of each maneuvering behavior to obtain the driver's behavioral characteristics The identification results are used to verify the credibility of the previously established identification model of driver behavior characteristics. 7. the verification of driver's behavior characteristic identification model prediction ability according to claim 1, it is characterized in that, select some drivers whose behavior characteristics are unknown to carry out steering, braking, acceleration test on the real vehicle driving simulator test platform, Collect the driver's manipulation behavior signal and the vehicle's motion state, and extract the characteristic parameters that can characterize the driver's steering, braking, and acceleration behavior characteristics, and send them into the characteristic identification model of each manipulation behavior to obtain the driver's behavior characteristics identification results.