Method, device, equipment, medium and product for recognizing driving perception based on electroencephalogram signals

Abstract

The application provides a method, device, equipment, medium and product for recognizing driving perception based on electroencephalogram signals, and relates to the technical field of data processing. The method comprises the following steps: collecting initial sample electroencephalogram signals generated by a driver; extracting intermediate sample electroencephalogram signals from the initial sample electroencephalogram signals; determining first sample electroencephalogram signals of the driver in a braking response stage and first sample electroencephalogram signals of the driver in a braking perception stage from the intermediate sample electroencephalogram signals based on braking markers; performing data enhancement on second sample electroencephalogram signals to obtain sample electroencephalogram signals; inputting the sample electroencephalogram signals into a driving perception classification model to be trained to obtain sample driving perception stages; and updating the driving perception model to be trained based on sample driving perception stages corresponding to the sample electroencephalogram signals and corresponding labels until a trained driving perception classification model is obtained, so as to be used for monitoring a dangerous perception state of the driver in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a method, device, equipment, medium and product for recognizing driving perception based on electroencephalogram signals. BACKGROUND

[0002] With the rapid development of urbanization and the continuous progress of automatic driving technology, mixed traffic scenarios involving vehicles with different intelligence levels are becoming increasingly common. In recent years, the use of human factor driving data, especially physiological signals such as electroencephalogram (EEG) to monitor the state of drivers has made a significant contribution to traffic safety, traffic planning and vehicle intelligence. The processing technology of human factor driving data can provide in-depth understanding of the behavior of drivers during driving, and provide technical support for improving traffic safety and efficiency.

[0003] At present, the related technologies are mostly concentrated in the use of electroencephalogram signals (EEG data) to monitor and evaluate the driving behavior of drivers, such as acceleration, deceleration, turning, lane changing and other driving behaviors, which cannot monitor the ability of drivers to perceive potential dangers. However, the perception and reaction ability of drivers to danger plays a crucial role in dealing with mixed traffic scenarios involving vehicles with different intelligence levels. Therefore, how to monitor the state of drivers' risk perception in real time, improve traffic safety and reduce accidents has important practical significance for human society. SUMMARY

[0004] In view of the above technical problems, the present application provides a method, device, equipment, medium and product for recognizing driving perception based on electroencephalogram signals, which aims to effectively monitor and evaluate the state of drivers' risk perception through electroencephalogram signals.

[0005] The first aspect of the present application provides a method for recognizing driving perception based on electroencephalogram signals, which comprises:

[0006] Collecting initial sample electroencephalogram signals generated by drivers in simulating a plurality of driving simulation scenarios in a driving simulation platform;

[0007] Extracting intermediate sample electroencephalogram signals corresponding to a target period from the initial sample electroencephalogram signals, wherein the target period is a period consisting of a fixed time length before and after the time when the braking marker in the driving simulation scenario is placed as a reference;

[0008] Based on the braking mark, determining from the intermediate sample EEG signals a first sample EEG signal indicating that the driver is in a braking response phase and a first sample EEG signal indicating that the driver is in a braking perception phase; the first sample EEG signal in the braking response phase carries a label indicating that the driver is in a danger response phase, and the first sample EEG signal in the braking perception phase carries a label indicating that the driver is in a danger perception phase;

[0009] performing data enhancement on the second sample EEG signals to obtain sample EEG signals to balance the number of the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase; the second sample EEG signals being the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase, whichever are smaller;

[0010] inputting the sample EEG signals into a driving perception classification model to be trained to obtain a sample driving perception stage corresponding to the sample EEG signals, and updating the driving perception model to be trained based on the sample driving perception stage corresponding to the sample EEG signals and a label corresponding to the sample EEG signals, until a trained driving perception classification model is obtained;

[0011] The EEG signal to be tested is input into the trained driving perception classification model to obtain the driving perception stage of the driver corresponding to the EEG signal to be tested, and the driving perception stage indicates that the driver is in a danger response stage or a danger perception stage.

[0012] A second aspect of the present invention provides a device for identifying driving perception based on EEG signals, the device comprising:

[0013] A signal acquisition module is used to collect initial sample EEG signals generated by the driver simulating multiple driving simulation scenarios in the driving simulation platform;

[0014] a signal extraction module for extracting intermediate sample EEG signals corresponding to a target period from the initial sample EEG signals, wherein the target period is a period consisting of a fixed duration before and after the moment of the braking mark in the driving simulation scene;

[0015] a signal differentiation module, configured to determine, based on the braking mark, from the intermediate sample EEG signals a first sample EEG signal indicating that the driver is in a braking response phase and a first sample EEG signal indicating that the driver is in a braking perception phase; the first sample EEG signal in the braking response phase carries a label indicating that the driver is in a danger response phase, and the first sample EEG signal in the braking perception phase carries a label indicating that the driver is in a danger perception phase;

[0016] a data enhancement module, configured to perform data enhancement on the second sample EEG signals to obtain sample EEG signals to balance the number of the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase; the second sample EEG signals being the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase, whichever are smaller;

[0017] a model training module, configured to input the sample EEG signals into a driving perception classification model to be trained, obtain a sample driving perception stage corresponding to the sample EEG signals, and update the driving perception model to be trained based on the sample driving perception stage corresponding to the sample EEG signals and a label corresponding to the sample EEG signals, until a trained driving perception classification model is obtained;

[0018] The model application module is used to input the EEG signal to be tested into the trained driving perception classification model to obtain the driving perception stage of the driver corresponding to the EEG signal to be tested, and the driving perception stage indicates that the driver is in a danger response stage or a danger perception stage.

[0019] The third aspect of the present invention provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the computer program is executed by the processor, it implements the method for recognizing driving perception based on EEG signals as described in the first aspect of the embodiment of the present invention.

[0020] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for recognizing driving perception based on EEG signals according to the first aspect of the embodiment of the present invention is implemented.

[0021] The fifth aspect of the present invention provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method for identifying driving perception based on EEG signals according to the first aspect of the embodiment of the present invention.

[0022] In the method for identifying driving perception based on EEG signals provided by the present invention, initial sample EEG signals generated by a driver in a driving simulation scenario are collected, and intermediate sample EEG signals corresponding to a target period are extracted from them. Based on braking marks in the driving simulation scenario, first sample EEG signals of the driver in a braking response phase and first sample EEG signals of the driver in a braking perception phase are determined from the intermediate sample EEG signals. Data enhancement is performed on the small number of first sample EEG signals to obtain sample EEG signals for model training. The sample EEG signals are then input into a driving perception classification model to be trained to obtain a sample driving perception phase output by the model. The driving perception model to be trained is trained based on the sample driving perception phase corresponding to the sample EEG signals and the phase labels corresponding to the sample EEG signals to obtain a trained driving perception classification model. In this way, the trained driving perception classification model in this embodiment can predict the driver's danger perception phase and danger reaction phase, thereby enabling real-time monitoring of the driver's danger perception state through the driving perception classification model, providing important support for improving traffic safety, optimizing traffic planning, and promoting vehicle intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0024] Figure 1 This is a flowchart of a method for identifying driving perception based on EEG signals according to an embodiment of the present invention;

[0025] Figure 2 This is a schematic diagram of a simulation of a vehicle approaching from the side, showing an embodiment of the present invention;

[0026] Figure 3 This is an overall framework diagram of a driving perception classification model according to an embodiment of the present invention;

[0027] Figure 4 1 is a schematic diagram of confusion matrix results of a test set before and after data augmentation processing, according to an embodiment of the present invention;

[0028] Figure 5 This is a structural block diagram of a device for recognizing driving perception based on EEG signals provided by one embodiment of the present invention;

[0029] Figure 6 FIG. 1 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0031] Please refer to Figure 1 , Figure 1 is a step flow chart of a method for recognizing driving perception based on electroencephalogram signals according to an embodiment of the present application. As shown in Figure 1 , the method for recognizing driving perception based on electroencephalogram signals provided by the present embodiment at least includes the following steps:

[0032] Step S11: Collecting initial sample electroencephalogram signals generated by a driver in simulating a plurality of driving simulation scenarios in a driving simulation platform.

[0033] In the present embodiment, a plurality of driving simulation scenarios related to braking events are simulated in the driving simulation platform, and drivers are invited to conduct simulation tests in the driving simulation platform. Electroencephalogram signals (EEG signals) generated by the drivers in simulating a plurality of driving simulation scenarios in the driving simulation platform are collected as initial sample electroencephalogram signals, which are used as original sample data for training a driving perception classification model. The electroencephalogram signal is an important biological signal used to record and analyze the electrical activity of the human brain.

[0034] Among them, the driver is not limited to male or female, and the selection criteria of the driver include but are not limited to: the corrected visual acuity of all participants is not less than 5.0, holding a C1 or C2 license, having certain off-site driving experience, having no other mental or physical diseases that may affect the test, being physically healthy, and having sufficient sleep in the recent period. Before participating in the test, the participants are fully informed of the experimental procedures and precautions, and it is ensured that the experiment will not harm human health.

[0035] Step S12: Extracting intermediate sample electroencephalogram signals corresponding to a target period from the initial sample electroencephalogram signals, the target period being a period composed of a fixed time length before and after the time point of the braking marker in the driving simulation scenario.

[0036] In this embodiment, corresponding braking markers are pre-assigned for multiple driving simulation scenarios. These braking markers represent the driver's latest braking action (e.g., braking maneuvers) in response to hazards in the driving simulation scenarios. Because the time period corresponding to the initial sample EEG signals is relatively long and not all data is valid for training the driving perception classification model, this embodiment determines a target epoch based on a certain time period before and after the braking marker. The target epoch divides long continuous data into shorter time periods for time series analysis and processing. Specifically, the target epoch in this embodiment is a epoch consisting of a fixed time period before and after the braking marker in the driving simulation scenario. It should be noted that the fixed time periods before and after the braking marker can be the same or different, and this is not a limitation. In one alternative example, the target epoch is defined as the period from 0.5 seconds before to 1.5 seconds after the braking marker, with a duration of 2 seconds.

[0037] In this way, this embodiment can extract the intermediate sample EEG signal corresponding to the target period from the initial sample EEG signal. For example, the epoch extraction function in the MNE-Python tool library is used to extract the intermediate sample EEG signal corresponding to the target period from the initial sample EEG signal. At this time, the intermediate sample EEG signal is the EEG signal for the driver to effectively perceive and respond to the danger in the driving simulation scenario. The intermediate sample EEG signal helps to capture the short-term changes and dynamic characteristics in the data.

[0038] Step S13: Based on the braking mark, determine the first sample EEG signal of the driver in the braking response stage and the first sample EEG signal of the driver in the braking perception stage from the intermediate sample EEG signals; the label carried by the first sample EEG signal in the braking response stage is that the driver is in the danger response stage, and the label carried by the first sample EEG signal in the braking perception stage is that the driver is in the danger perception stage.

[0039] In this embodiment, after obtaining the intermediate sample EEG signals, the first sample EEG signals indicating that the driver is in the braking response phase and the first sample EEG signals indicating that the driver is in the braking perception phase can be determined from the intermediate sample EEG signals based on the braking marker. The braking response phase can be the phase in which the driver responds to danger and takes braking action, while the braking perception phase can be the phase in which the driver senses danger but has not yet responded. The first sample EEG signal in the braking response phase carries the label "driver is in the danger response phase," and the first sample EEG signal in the braking perception phase carries the label "driver is in the danger perception phase."

[0040] Unlike related technologies that focus solely on second-level hazard perception, this embodiment focuses on analyzing first- and second-level hazard perception through EEG signals. This approach uses EEG signals to monitor and evaluate the driver's risk perception at various stages of a dangerous scenario, enabling real-time monitoring of the driver's hazard perception state and classifying the driver's hazard perception status during driving. First-level hazard perception refers to the driver's alertness to potential danger in the current environment, while second-level hazard perception (hazard response) refers to the driver's reaction to a potential danger.

[0041] Step S14: Data enhancement is performed on the second sample EEG signal to obtain a sample EEG signal to balance the number of the first sample EEG signals in the braking response stage and the first sample EEG signals in the braking perception stage; the second sample EEG signal is the first sample EEG signal with a smaller number between the first sample EEG signals in the braking response stage and the first sample EEG signals in the braking perception stage.

[0042] Since EEG signals in the braking perception stage are more difficult to capture, it is impossible to obtain data corresponding to the current driver's danger perception process through obvious labels. Therefore, some of the collected data may be invalid. This leads to an imbalance in the number of the first-sample EEG signals in the braking response stage and the first-sample EEG signals in the braking perception stage in the data set of the first-sample EEG signals collected. This imbalance may lead to overfitting problems if the model is directly trained.

[0043] Therefore, this embodiment determines the first sample EEG signal of the braking response phase and the first sample EEG signal of the braking perception phase as the second sample EEG signal, and performs data enhancement on the second sample EEG signal to obtain a sample EEG signal so that the number of the first sample EEG signals of the braking response phase and the first sample EEG signals of the braking perception phase are balanced. The sample EEG signals include: the first sample EEG signals of the braking response phase and the first sample EEG signals of the braking perception phase that are balanced in number. It can be understood that each sample EEG signal after data enhancement in this embodiment carries a corresponding label.

[0044] Step S15: Input the sample EEG signal into the driving perception classification model to be trained to obtain the sample driving perception stage corresponding to the sample EEG signal, and update the driving perception model to be trained based on the sample driving perception stage corresponding to the sample EEG signal and the label corresponding to the sample EEG signal until a trained driving perception classification model is obtained.

[0045] In this embodiment, after obtaining the sample EEG signal, the sample EEG signal can be input into the driving perception classification model to be trained. Through processing by the driving perception classification model to be trained, the output of the driving perception classification model to be trained is obtained: the sample driving perception stage corresponding to the sample EEG signal. The sample driving perception stage is the model output of the model training stage, indicating that the driver corresponding to the sample EEG signal is in the danger response stage, or in the danger perception stage.

[0046] After obtaining the sample driving perception stage corresponding to the sample EEG signal, the model parameters of the driving perception model to be trained can be updated based on the sample driving perception stage corresponding to the sample EEG signal and the label corresponding to the sample EEG signal until a trained driving perception classification model is obtained. In this embodiment, the trained driving perception classification model can classify the driver's danger perception stage during driving based on the EEG signal: determine whether the driver is in the danger response stage or the danger perception stage.

[0047] Step S16: Inputting the EEG signal to be measured into the trained driving perception classification model to obtain the driving perception stage of the driver corresponding to the EEG signal to be measured, wherein the driving perception stage indicates that the driver is in a danger response stage or a danger perception stage.

[0048] In this embodiment, the EEG signal to be measured is an EEG signal that needs to be classified for driving perception. The EEG signal to be measured can be a real-time EEG signal collected by the driver during the driving process. The driving perception classification model of this embodiment is used to automatically identify the driver's danger perception state based on the EEG signal to be measured, and determine whether the driver is in the danger response stage or the danger perception stage, thereby improving traffic safety and reducing the occurrence of accidents, which has important practical significance for human society.

[0049] The collected EEG signal to be tested can be input into a trained driving perception classification model to obtain the driver's driving perception stage corresponding to the EEG signal to be tested, as output by the trained driving perception classification model. The predicted driving perception stage indicates that the driver is in a dangerous response stage or a dangerous perception stage. The driving perception stage represents the driving perception state, the dangerous response stage indicates that the driver is in a dangerous response state, and the dangerous perception stage indicates that the driver is in a dangerous perception stage.

[0050] In this embodiment, the trained driving perception classification model can predict the driver's danger perception stage and danger response stage, thereby realizing real-time monitoring of the driver's danger perception state through the driving perception classification model, identifying the driver's perception of environmental hazards in different driving stages, and realizing automatic identification of the driver's danger perception state, providing important support for improving traffic safety, optimizing traffic planning and promoting vehicle intelligence.

[0051] In combination with the above embodiments, in one embodiment, the present invention further provides a method for identifying driving perception based on EEG signals. In this method, in addition to the above steps, steps S21 and S22 may be further included. Furthermore, the above step S13 of "determining, based on the braking mark, the first sample EEG signal of the driver in the braking response stage and the first sample EEG signal of the driver in the braking perception stage from the intermediate sample EEG signals" may specifically include steps S23 and S24:

[0052] Step S21: establishing a plurality of driving simulation scenarios, wherein the plurality of driving simulation scenarios include at least: a driving simulation scenario of an emergency braking of a vehicle ahead, a driving simulation scenario of pedestrians crossing the road, and a driving simulation scenario of a sudden lane change of a side vehicle.

[0053] In this embodiment, multiple driving simulation scenarios are established, and different driving simulation scenarios correspond to different deceleration (braking) trigger conditions. The multiple driving simulation scenarios include at least: a driving simulation scenario in which the vehicle ahead urgently brakes, a driving simulation scenario in which pedestrians cross the road, and a driving simulation scenario in which a side vehicle suddenly changes lanes.

[0054] In the driving simulation scenario of a leading vehicle suddenly braking, the driver simulates an emergency braking situation: the host vehicle (i.e., the test vehicle) follows the leading vehicle, which maintains the same speed as the host vehicle. However, after traveling a few meters (e.g., 50 meters), the leading vehicle suddenly brakes, rapidly dropping to zero speed. The driver must quickly decelerate to avoid a rear-end collision when approaching this situation.

[0055] In the driving simulation scenario involving a sudden lane change by a lateral vehicle, the simulation simulates a sudden lane change. For example, if a preceding lateral vehicle suddenly changes lanes, while the host vehicle is traveling in its designated lane, the preceding lateral vehicle will travel to a fixed position at 0.7 times the host vehicle's speed and, after a certain number of meters (e.g., 50 meters), make an emergency lane change back to the host vehicle's lane. Upon noticing this change, the driver must slow down to avoid a rear-end collision. Of course, the simulation can also simulate a sudden lane change by a following lateral vehicle at an accelerated speed, which is not a limitation of the present invention.

[0056] In the pedestrian crossing driving simulation scenario, the driver simulates a pedestrian crossing the road: when the host vehicle arrives at a fixed position, a pedestrian crosses the road in front of the host vehicle at a certain speed (for example, 5 meters per second). The driver must react quickly and slow down to avoid hitting the pedestrian.

[0057] Take the example of a vehicle changing lanes suddenly sideways. Figure 2 As shown, Figure 2 FIG. 1 is a schematic diagram showing a simulation of a vehicle entering from the side according to an embodiment of the present invention. Figure 2In the middle, the blue car is the main car, and the red car is the car entering from the side. Figure 2 As can be seen, this scenario goes through the normal driving phase, phase 1 (stage 1), and phase 2 (stage 2). Phase 1 is when the driver of the main vehicle senses a car driving sideways in the side lane and may change lanes, which is the danger perception phase. Phase 2 is when the driver of the main vehicle responds to the car changing lanes and brakes, which is the danger response phase.

[0058] It can be understood that, except for the sudden lane change of the lateral vehicle, in fact, the above-mentioned multiple driving simulation scenarios will all go through the normal driving stage, the danger perception stage and the danger response stage. At this time, it is necessary to set different braking marks in multiple driving simulation scenarios, and extract the first sample EEG signal of the driver in the braking response stage and the first sample EEG signal of the driver in the braking perception stage from the initial sample EEG signals in each driving simulation scenario.

[0059] Step S22: The moment of the brake mark in the driving simulation scenario of emergency braking of the vehicle in front is set to the moment when the vehicle in front decelerates to zero; the moment of the brake mark in the driving simulation scenario of pedestrians crossing the road is set to the moment when the pedestrians start to cross the road; the moment of the brake mark in the driving simulation scenario of a sudden lane change of a lateral vehicle is set to the moment when the lateral vehicle suddenly changes lanes.

[0060] In this embodiment, for the driving simulation scenario of the emergency braking of the vehicle in front, the moment of the braking mark of the driving simulation scenario of the emergency braking of the vehicle in front can be set to the moment when the vehicle in front decelerates to zero; for the driving simulation scenario of pedestrians crossing the road, the moment of the braking mark of the driving simulation scenario of pedestrians crossing the road can be set to the moment when the pedestrians start to cross the road; for the driving simulation scenario of the sudden lane change of the side vehicle, the moment of the braking mark of the driving simulation scenario of the sudden lane change of the side vehicle can be set to the moment when the side vehicle suddenly changes lanes.

[0061] Step S23: determining, among the intermediate sample EEG signals, the intermediate sample EEG signals before the moment of the braking mark as the first sample EEG signals of the driver in the braking perception stage.

[0062] In this embodiment, since the braking mark of this embodiment is the latest response of the driver to the danger in the driving simulation scenario to generate braking action, it is considered that before the moment when the braking mark is located, the driver is still in the stage of discovering the danger but not responding. At this time, the intermediate sample EEG signal before the moment when the braking mark is located can be determined as the first sample EEG signal of the driver in the braking perception stage.

[0063] Step S24: determining, among the intermediate sample EEG signals, the intermediate sample EEG signals that are after the moment of the braking mark as the first sample EEG signals of the driver in the braking response phase.

[0064] In this embodiment, it can be considered that after the moment when the brake mark is located, the driver must respond to avoid danger. Therefore, the intermediate sample EEG signal after the moment when the brake mark is located in the intermediate sample EEG signal can be determined as the first sample EEG signal of the driver in the braking response stage.

[0065] In combination with the above embodiments, in one embodiment, the present invention further provides a method for identifying driving perception based on EEG signals. In this method, the "inputting the sample EEG signal into the driving perception classification model to be trained to obtain the sample driving perception stage corresponding to the sample EEG signal" in the above step S15 may specifically include steps S31 to S34:

[0066] Step S31: inputting the sample EEG signal into the two-dimensional convolution filter to obtain a plurality of F1 feature maps; the plurality of F1 feature maps correspond to sample EEG signals of different bandpass frequencies.

[0067] In this embodiment, the driving perception classification model to be trained includes at least a two-dimensional convolution filter, a depthwise convolution module, and a separable convolution module. When a sample EEG signal is input into the driving perception classification model to be trained, the first convolution layer of the model is a two-dimensional convolution filter. The sample EEG signal is first input into the two-dimensional convolution filter, which then processes the sample EEG signal to produce multiple F1 feature maps output by the two-dimensional convolution filter. These F1 feature maps correspond to sample EEG signals with different bandpass frequencies.

[0068] Step S32: Process each of the multiple F1 feature maps through the depth convolution module, and control the number of spatial filters learned by each F1 feature map by depth to obtain a first feature map.

[0069] In this embodiment, the second convolutional layer of the driving perception classification model to be trained is a deep convolution module. After obtaining multiple F1 feature maps output by the two-dimensional convolution filter, each of these F1 feature maps is processed by the deep convolution module. The depth (i.e., parameter D) controls the number of spatial filters to be learned for each F1 feature map, resulting in the first feature map output by the deep convolution module. The kernel size of the deep convolution module in this embodiment is (C, 1). Unlike traditional convolutional neural networks, this deep convolution module operates on each feature map rather than fully connecting all previous feature maps.

[0070] Step S33: Process the first feature map through a separable convolution module to obtain a second feature map.

[0071] In this embodiment, the third convolution layer of the driving perception classification model to be trained is a separable convolution module, and the first feature map can be processed by the separable convolution module to obtain a second feature map output by the separable convolution module.

[0072] Step S34: performing batch normalization on the second feature map along the feature map dimension to obtain the sample driving perception stage corresponding to the sample EEG signal.

[0073] In this embodiment, after obtaining the second feature map, the second feature map can be batch normalized along the feature map dimension to obtain the sample driving perception stage corresponding to the sample EEG signal output by the driving perception classification model to be trained.

[0074] In combination with the above embodiments, in one embodiment, the present invention further provides a method for identifying driving perception based on EEG signals. In this embodiment, the convolution kernel length of the two-dimensional convolution filter is half the data sampling rate corresponding to the sample EEG signal. The driving perception classification model to be trained further includes: a pooling layer with an average size of (1, 4); the above step S34 may specifically include steps S41 and S42:

[0075] Step S41: performing batch normalization on the second feature map along the feature map dimension to obtain a third feature map.

[0076] In this embodiment, before using nonlinear activation, the model can first apply batch normalization along the feature map dimension, that is, batch normalization is performed on the second feature map, and random dropout technology is used to prevent overfitting to obtain a third feature map.

[0077] Step S42: Using the pooling layer to process the third feature map to reduce the data sampling rate corresponding to the sample EEG signal, and obtain the sample driving perception stage corresponding to the sample EEG signal.

[0078] In this embodiment, after obtaining the third feature map, a pooling layer with an average size of (1, 4) is used to process the third feature map to reduce the data sampling rate corresponding to the sample EEG signal, and a maximum norm constraint of 1 is used to normalize the weight of each spatial filter to obtain the sample driving perception stage corresponding to the sample EEG signal.

[0079] In an optional embodiment, binary cross entropy can be selected as the loss function of the driving perception classification model to be trained, as shown below:

[0080] ;

[0081] in, is the output of the driving perception classification model to be trained, is the label corresponding to the sample EEG signal. The driving perception model to be trained can be updated based on this loss function until the loss function converges, thus obtaining a trained driving perception classification model.

[0082] Furthermore, in one specific embodiment, the Adam optimization algorithm can be used to update network parameters. The initial learning rate is set to 0.001 to optimize model performance. Regarding parameter settings, the driving perception classification model in this embodiment can be a binary classification model with 59 channels. The dropout rate is set to 0.5, the convolution kernel length is 32, F1 is 8, D is 2, F2 is 16, the batch size is 16, and 1000 iterations are performed.

[0083] The driving perception classification model provided in this embodiment has a powerful feature extraction capability for EEG signals. Compared with previous methods that rely on manual feature extraction, the deep learning-based method can learn complex high-dimensional features that humans cannot understand through large amounts of data, thereby improving the classification accuracy and generalization performance of the model through parameter adjustment.

[0084] In one embodiment, Figure 3 As shown, Figure 3 This is an overall framework diagram of a driving perception classification model shown in one embodiment of the present invention. Figure 3 In

[15] , the driving behavior classification model includes at least: a two-dimensional convolution filter (conv2d), a depthwise convolution module (depthwise_conv2d), and a separable convolution module (separable_conv2d).

[0085] In an optional embodiment, the collected intermediate sample EEG signals can be divided (e.g., a total of 2391 target cycles of intermediate sample EEG signals from 30 subjects are collected), half of the data (i.e., 1196 intermediate sample EEG signals) are designated as a training set, one quarter (i.e., 597 intermediate sample EEG signals) are allocated as validation data, and the remaining one quarter (i.e., 598 intermediate sample EEG signals) are used for testing purposes.

[0086] In combination with the above embodiments, in one embodiment, the present invention further provides a method for identifying driving perception based on EEG signals. In this embodiment, the "performing data enhancement on the second sample EEG signal to obtain a sample EEG signal" in the above step S14 may specifically include steps S51 to S53:

[0087] Step S51: determining k adjacent sample EEG signals of the second sample EEG signal.

[0088] In the embodiment, the second sample EEG signals of the training set and the test set can be subjected to data enhancement by SMOTE oversampling. Specifically, k adjacent sample EEG signals of the second sample EEG signal can be determined first, where k is a natural number greater than 0. Generally, k = 5.

[0089] wherein the adjacent sample EEG signal is a sample EEG signal with a Euclidean distance from the second sample EEG signal lower than a threshold value. That is, the Euclidean distance between the sample EEG signal (sample) and the second sample EEG signal can be calculated, and according to the threshold value, the sample EEG signal with a Euclidean distance lower than the threshold value is regarded as the nearest neighbor sample, i.e., the adjacent sample EEG signal.

[0090] In an optional embodiment, generally, the number of the first sample EEG signals in the braking perception stage is less than the number of the first sample EEG signals in the braking response stage.

[0091] Step S52: randomly selecting one target adjacent sample EEG signal from the k adjacent sample EEG signals, and calculating the difference between the distance of the second sample EEG signal and the target adjacent sample EEG signal.

[0092] In the embodiment, one adjacent sample EEG signal from the k adjacent sample EEG signals is randomly selected as the target adjacent sample EEG signal, and the difference diff between the distance of the second sample EEG signal and the target adjacent sample EEG signal is calculated.

[0093] Step S53: based on the difference and a random number, generating a new synthetic sample EEG signal between the second sample EEG signal and the target adjacent sample EEG signal, until the number of the generated new synthetic sample EEG signals makes the number of the first sample EEG signals in the braking response stage and the number of the first sample EEG signals in the braking perception stage balanced, to obtain the sample EEG signals composed of the new synthetic sample EEG signals and the first sample EEG signals.

[0094] In the embodiment, after obtaining the difference diff, a new synthetic sample EEG signal can be generated between the second sample EEG signal and the target adjacent sample EEG signal based on the difference diff and a random number r between 0 and 1. The new synthetic sample EEG signal and the second sample EEG signal are first sample EEG signals belonging to the same stage, such as first sample EEG signals belonging to the braking response stage or first sample EEG signals belonging to the braking perception stage.

[0095] wherein the specific formula is as follows:

[0096] ;

[0097] in, To synthesize sample EEG signals, is the second sample EEG signal.

[0098] In this embodiment, the steps of randomly selecting a target adjacent sample EEG signal from the k adjacent sample EEG signals, and calculating the difference between the distances between the second sample EEG signal and the target adjacent sample EEG signal can be repeated, to the step of generating a new synthetic sample EEG signal between the second sample EEG signal and the target adjacent sample EEG signal based on the difference and the random number, until the number of generated new synthetic sample EEG signals balances the number of first sample EEG signals in the braking response stage and the number of first sample EEG signals in the braking perception stage, thereby obtaining a sample EEG signal, which includes: the new synthetic sample EEG signal and the first sample EEG signal before data enhancement (i.e., the first sample EEG signal in the braking response stage and the first sample EEG signal in the braking perception stage).

[0099] The data enhancement method proposed in this embodiment is used to upsample the data, generate approximate data points near the second sample EEG signal, and increase the sample size to solve problems such as unbalanced data distribution.

[0100] like Figure 4 As shown, Figure 4 FIG1 is a schematic diagram of the confusion matrix results of a test set before and after data enhancement processing according to an embodiment of the present invention. Figure 4 In the figure, the confusion matrix results of the test set before data augmentation are shown on the left, and the confusion matrix results of the test set after data augmentation are shown on the right. Stage 0 (stage0) and stage 1 (stage1) are different driving perception stages. The horizontal axis is the predicted result (Predicted label), and the vertical axis is the true label (True label). Figure 4 As can be seen on the left, the classification results are biased towards the danger response stage. Although the accuracy is not low in this case, it is suboptimal because many EEG signals in the danger perception stage are inaccurately classified as the danger response stage, which fails to classify correctly. Attempts to adjust the weight values ​​of the two stages also fail to alleviate the overfitting problem. Figure 4 As can be seen on the right side of , through the data enhancement method proposed in this embodiment, the classification results can effectively identify the danger perception stage and the danger response stage, and provide more reliable evaluation indicators.

[0101] In combination with the above embodiments, in one embodiment, the present invention further provides a method for identifying driving perception based on EEG signals. In this method, in addition to the step S12 of "extracting intermediate sample EEG signals corresponding to the target period from the initial sample EEG signals", the method may further include step S61, and the step S12 of "extracting intermediate sample EEG signals corresponding to the target period from the initial sample EEG signals" may specifically include step S62:

[0102] Step S61: The initial sample EEG signal is subjected to high-frequency noise filtering to obtain a third sample EEG signal having a frequency range lower than a filtering threshold.

[0103] In this embodiment, after obtaining the initial sample EEG signal, the initial sample EEG signal can be subjected to a high-frequency noise filter to obtain a third sample EEG signal having a frequency range below the filtering threshold (e.g., 40 Hz). The third sample EEG signal is a clean EEG signal. For example, the initial sample EEG signal can be subjected to a high-frequency noise filter using a linear phase FIR filter from 0.5 Hz to 40 Hz to eliminate unwanted frequency components and noise, retain the frequency range of interest, reduce noise interference, and improve overall signal quality.

[0104] Step S62: extracting the intermediate sample EEG signal corresponding to the target period from the third sample EEG signal.

[0105] In this embodiment, after the third sample EEG signal is obtained, an intermediate sample EEG signal corresponding to the target period is extracted from the third sample EEG signal.

[0106] In another optional embodiment, after obtaining the third sample EEG signal, the sampling frequency can be reduced (for example, from 1000Hz to 500Hz through direct sampling at equal intervals). This helps reduce the amount of data, lowers computational complexity, and accommodates the lower sampling rate requirements of subsequent analysis methods. Next, because the EEG signal may include eye movement signals, independent component analysis (ICA) is used to separate the mixed third sample EEG signal, retaining only the EEG, to obtain the fourth sample EEG signal. The purpose of ICA is to separate the different source signals (such as the EEG and eye movement) in the EEG data. Using ICA, for example, a 64-channel EEG signal can be converted into a 64-dimensional signal. One dimension of this signal may be the separated eye movement signal. Therefore, by deleting the signals from certain dimensions, the eye movement signal can be removed, facilitating more accurate subsequent analysis and interpretation. Finally, the intermediate sample EEG signal corresponding to the target period is extracted from the fourth sample EEG signal.

[0107] In summary, the method for identifying driving perception based on EEG signals provides an approach for classifying EEG data and comprehensively analyzing and identifying human-driven data. This method addresses the classification problem of the hazard perception stage during driving. It also employs a deep learning approach based on convolutional neural networks, which boasts powerful feature extraction capabilities for EEG signals. Compared to previous methods that rely on manual feature extraction, deep learning-based approaches can learn complex, high-dimensional features that are incomprehensible to humans from large amounts of data, thereby improving the model's classification accuracy and generalization performance through parameter adjustment. In this method, EEG signals in the experimental scenario are filtered to remove high-frequency noise, resulting in clean EEG signals with a frequency range below 40Hz. A data augmentation algorithm is then used to upsample the data, generating approximate data points near existing samples and increasing the sample size to address issues such as data imbalance. Finally, this data is input into the network to establish a driving perception classification model capable of identifying the driver's perception of environmental hazards at different driving stages. This model automatically identifies the driver's hazard perception state and predicts the driver's hazard perception and response stage. This method demonstrates its effectiveness in addressing the challenge of classifying human-driven data in mixed traffic scenarios. This method can also be further used to monitor and evaluate drivers' risk perception at various stages of dangerous scenarios, providing important support for improving traffic safety, optimizing traffic planning, and advancing vehicle intelligence.

[0108] It should be noted that for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0109] Based on the same inventive concept, an embodiment of the present invention provides a device for recognizing driving perception based on EEG signals. Figure 5 , Figure 5 This is a structural block diagram of a device for recognizing driving perception based on EEG signals provided by one embodiment of the present invention. Figure 5 As shown, the device for recognizing driving perception based on EEG signals in this embodiment may include:

[0110] A signal acquisition module is used to collect initial sample EEG signals generated by the driver simulating multiple driving simulation scenarios in the driving simulation platform;

[0111] a signal extraction module configured to extract an intermediate sample electroencephalogram signal corresponding to a target period from the initial sample electroencephalogram signal, the target period being a period formed by a fixed time length before and after a time point of a brake marker in the driving simulation scenario;

[0112] a signal distinguishing module configured to determine, based on the brake marker, a first sample electroencephalogram signal of a driver in a brake response stage and a first sample electroencephalogram signal of a driver in a brake perception stage from the intermediate sample electroencephalogram signal; the first sample electroencephalogram signal of the brake response stage carries a label of a driver in a dangerous response stage, and the first sample electroencephalogram signal of the brake perception stage carries a label of a driver in a dangerous perception stage;

[0113] a data enhancement module configured to perform data enhancement on a second sample electroencephalogram signal to obtain a sample electroencephalogram signal, so as to balance the number of the first sample electroencephalogram signal of the brake response stage and the first sample electroencephalogram signal of the brake perception stage; the second sample electroencephalogram signal is the first sample electroencephalogram signal with a smaller number in the first sample electroencephalogram signal of the brake response stage and the first sample electroencephalogram signal of the brake perception stage;

[0114] a model training module configured to input the sample electroencephalogram signal into a driving perception classification model to be trained, to obtain a sample driving perception stage corresponding to the sample electroencephalogram signal, and to update the driving perception model to be trained based on the sample driving perception stage corresponding to the sample electroencephalogram signal and a label corresponding to the sample electroencephalogram signal, until a trained driving perception classification model is obtained;

[0115] a model application module configured to input a to-be-tested electroencephalogram signal into the trained driving perception classification model to obtain a driving perception stage of a driver corresponding to the to-be-tested electroencephalogram signal, the driving perception stage indicating that the driver is in a dangerous response stage or a dangerous perception stage.

[0116] Optionally, the apparatus further comprises:

[0117] a scenario establishing module configured to establish a plurality of driving simulation scenarios, the plurality of driving simulation scenarios at least including a driving simulation scenario of an emergency braking of a front vehicle, a driving simulation scenario of a pedestrian crossing a road, and a driving simulation scenario of a sharp lane change of a lateral vehicle;

[0118] a marker setting module configured to set a time point of a brake marker in the driving simulation scenario of the emergency braking of the front vehicle as a time point at which the front vehicle is decelerated to zero, set a time point of a brake marker in the driving simulation scenario of the pedestrian crossing the road as a time point at which the pedestrian starts to cross the road, and set a time point of a brake marker in the driving simulation scenario of the sharp lane change of the lateral vehicle as a time point at which the lateral vehicle sharply changes lanes.

[0119] The signal distinguishing module includes:

[0120] a first determining module, configured to determine, among the intermediate sample EEG signals, an intermediate sample EEG signal that is before the moment of the braking mark as a first sample EEG signal of the driver in a braking perception stage;

[0121] The second determining module is configured to determine, among the intermediate sample EEG signals, the intermediate sample EEG signals that are located after the moment of the braking mark as the first sample EEG signals of the driver in the braking response phase.

[0122] Optionally, the driving perception classification model to be trained includes at least: a two-dimensional convolution filter, a depth convolution module, and a separable convolution module;

[0123] Model training module, including:

[0124] A first convolution module is configured to input the sample EEG signal into the two-dimensional convolution filter to obtain a plurality of F1 feature maps; the plurality of F1 feature maps correspond to sample EEG signals of different bandpass frequencies;

[0125] A second convolution module is used to process each of the multiple F1 feature maps through the depth convolution module, and control the number of spatial filters learned by each F1 feature map by depth to obtain a first feature map;

[0126] a third convolution module, configured to process the first feature map through a separable convolution module to obtain a second feature map;

[0127] A normalization module is used to perform batch normalization on the second feature map along the feature map dimension to obtain a sample driving perception stage corresponding to the sample EEG signal.

[0128] Optionally, the convolution kernel length of the two-dimensional convolution filter is half of the data sampling rate corresponding to the sample EEG signal, and the driving perception classification model to be trained further includes: a pooling layer with an average size of (1, 4);

[0129] The normalization module includes:

[0130] A batch processing module, configured to perform batch normalization on the second feature map along the feature map dimension to obtain a third feature map;

[0131] A pooling processing module is used to use the pooling layer to process the third feature map to reduce the data sampling rate corresponding to the sample EEG signal and obtain the sample driving perception stage corresponding to the sample EEG signal.

[0132] Optionally, the data enhancement module includes:

[0133] an adjacent sample determination module, configured to determine k adjacent sample EEG signals of the second sample EEG signal;

[0134] a difference determination module, configured to randomly select a target adjacent sample EEG signal from the k adjacent sample EEG signals, and calculate a difference between the distances of the second sample EEG signal and the target adjacent sample EEG signal;

[0135] A synthetic sample generation module is used to generate a new synthetic sample EEG signal between the second sample EEG signal and the target adjacent sample EEG signal based on the difference and the random number, until the number of new synthetic sample EEG signals generated balances the number of first sample EEG signals in the braking response phase and the number of first sample EEG signals in the braking perception phase, thereby obtaining the sample EEG signal composed of the new synthetic sample EEG signal and the first sample EEG signal.

[0136] Optionally, the device further comprises:

[0137] a filtering module configured to filter the initial sample EEG signal through a high-frequency noise filter before extracting the intermediate sample EEG signal corresponding to the target period from the initial sample EEG signal to obtain a third sample EEG signal having a frequency range lower than a filtering threshold;

[0138] The signal extraction module includes:

[0139] The signal extraction submodule is used to extract the intermediate sample EEG signal corresponding to the target period from the third sample EEG signal.

[0140] Based on the same inventive concept, another embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps in the method for identifying driving perception based on EEG signals as described in any of the above embodiments of the present invention.

[0141] Based on the same inventive concept, another embodiment of the present invention provides an electronic device, such as Figure 6 shown. Figure 6 This is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the steps of the method for identifying driving perception based on EEG signals described in any of the above embodiments of the present invention.

[0142] Based on the same inventive concept, another embodiment of the present application provides a computer program product comprising computer programs / instructions which, when executed by a processor, implement the steps of the method for recognizing driving perception based on electroencephalogram signals according to any one of the above embodiments of the present application.

[0143] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts are described in the part of the method embodiment.

[0144] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same and similar parts between embodiments can be referred to each other.

[0145] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a device, or a computer program product. Therefore, the embodiments of the present application can be in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] The embodiments of the present application are described with reference to flowcharts and / or block diagrams of the method, terminal device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal device produce a device implemented in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0147] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing terminal device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product comprising instruction apparatus, which implements the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0149] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0150] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0151] The above is a detailed introduction to the method, device, equipment, medium and product for recognizing driving perception based on EEG signals provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for identifying driving perception based on EEG signals, characterized in that: The method comprises: Collecting initial sample EEG signals generated by the driver simulating multiple driving simulation scenarios in the driving simulation platform; Extracting an intermediate sample EEG signal corresponding to a target period from the initial sample EEG signal, wherein the target period is a period consisting of a fixed time period before and after the moment of the braking mark in the driving simulation scene; Based on the braking mark, determining from the intermediate sample EEG signals a first sample EEG signal indicating that the driver is in a braking response phase and a first sample EEG signal indicating that the driver is in a braking perception phase; the first sample EEG signal in the braking response phase carries a label indicating that the driver is in a danger response phase, and the first sample EEG signal in the braking perception phase carries a label indicating that the driver is in a danger perception phase; performing data enhancement on the second sample EEG signals to obtain sample EEG signals to balance the number of the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase; the second sample EEG signals being the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase, whichever are smaller; inputting the sample EEG signals into a driving perception classification model to be trained to obtain a sample driving perception stage corresponding to the sample EEG signals, and updating the driving perception model to be trained based on the sample driving perception stage corresponding to the sample EEG signals and a label corresponding to the sample EEG signals, until a trained driving perception classification model is obtained; Inputting the EEG signal to be tested into the trained driving perception classification model to obtain the driving perception stage of the driver corresponding to the EEG signal to be tested, wherein the driving perception stage indicates that the driver is in a danger response stage or a danger perception stage; The driving perception classification model to be trained includes at least: a two-dimensional convolution filter, a depth convolution module, and a separable convolution module; Inputting the sample EEG signal into a driving perception classification model to be trained to obtain a sample driving perception stage corresponding to the sample EEG signal includes: Inputting the sample EEG signal into the two-dimensional convolution filter to obtain a plurality of F1 feature maps; the plurality of F1 feature maps correspond to sample EEG signals of different bandpass frequencies; Processing each of the multiple F1 feature maps by the depth convolution module, and controlling the number of spatial filters learned by each F1 feature map by depth, to obtain a first feature map; Processing the first feature map through a separable convolution module to obtain a second feature map; Batch normalization is performed on the second feature map along the feature map dimension to obtain a sample driving perception stage corresponding to the sample EEG signal.

2. The method for recognizing driving perception based on EEG signals according to claim 1, characterized in that: The method further comprises: Establishing multiple driving simulation scenarios, the multiple driving simulation scenarios at least including: a driving simulation scenario of an emergency braking of a vehicle ahead, a driving simulation scenario of a pedestrian crossing the road, and a driving simulation scenario of a sudden lane change of a side vehicle; The time when the brake mark of the driving simulation scene of the preceding vehicle emergency braking is set to the time when the preceding vehicle decelerates to zero; the time when the brake mark of the driving simulation scene of pedestrians crossing the road is set to the time when the pedestrians start crossing the road; the time when the brake mark of the driving simulation scene of the lateral vehicle suddenly changing lanes is set to the time when the lateral vehicle suddenly changes lanes; Determining, based on the braking mark, a first sample EEG signal of the driver in a braking response phase and a first sample EEG signal of the driver in a braking perception phase from the intermediate sample EEG signals, comprising: Determining, among the intermediate sample EEG signals, an intermediate sample EEG signal before the moment of the braking mark as a first sample EEG signal of the driver in the braking perception stage; Among the intermediate sample EEG signals, the intermediate sample EEG signals that are located after the moment of the braking mark are determined as the first sample EEG signals of the driver in the braking response phase.

3. The method for recognizing driving perception based on EEG signals according to claim 1, characterized in that: The convolution kernel length of the two-dimensional convolution filter is half of the data sampling rate corresponding to the sample EEG signal. The driving perception classification model to be trained further includes: a pooling layer with an average size of (1, 4); Batch normalizing the second feature map along the feature map dimension to obtain a sample driving perception stage corresponding to the sample EEG signal, including: performing batch normalization on the second feature map along the feature map dimension to obtain a third feature map; The third feature map is processed using the pooling layer to reduce a data sampling rate corresponding to the sample EEG signal, thereby obtaining a sample driving perception stage corresponding to the sample EEG signal.

4. The method for recognizing driving perception based on EEG signals according to any one of claims 1 to 3, characterized in that: Performing data enhancement on the second sample EEG signal to obtain a sample EEG signal, including: Determining k adjacent sample EEG signals of the second sample EEG signal; Randomly selecting a target adjacent sample EEG signal from the k adjacent sample EEG signals, and calculating a difference between the distances of the second sample EEG signal and the target adjacent sample EEG signal; Based on the difference and the random number, a new synthetic sample EEG signal is generated between the second sample EEG signal and the target adjacent sample EEG signal, until the number of generated new synthetic sample EEG signals balances the number of first sample EEG signals in the braking response phase and the number of first sample EEG signals in the braking perception phase, thereby obtaining the sample EEG signal composed of the new synthetic sample EEG signal and the first sample EEG signal.

5. The method for recognizing driving perception based on EEG signals according to any one of claims 1 to 3, characterized in that: Before extracting the intermediate sample EEG signal corresponding to the target period from the initial sample EEG signal, the method further includes: The initial sample EEG signal is subjected to high-frequency noise filtering to obtain a third sample EEG signal having a frequency range lower than a filtering threshold; Extracting an intermediate sample EEG signal corresponding to a target period from the initial sample EEG signal includes: An intermediate sample EEG signal corresponding to the target period is extracted from the third sample EEG signal.

6. A device for identifying driving perception based on EEG signals, characterized in that: The device comprises: A signal acquisition module is used to collect initial sample EEG signals generated by the driver simulating multiple driving simulation scenarios in the driving simulation platform; a signal extraction module for extracting intermediate sample EEG signals corresponding to a target period from the initial sample EEG signals, wherein the target period is a period consisting of a fixed duration before and after the moment of the braking mark in the driving simulation scene; a signal differentiation module, configured to determine, based on the braking mark, from the intermediate sample EEG signals a first sample EEG signal indicating that the driver is in a braking response phase and a first sample EEG signal indicating that the driver is in a braking perception phase; the first sample EEG signal in the braking response phase carries a label indicating that the driver is in a danger response phase, and the first sample EEG signal in the braking perception phase carries a label indicating that the driver is in a danger perception phase; a data enhancement module, configured to perform data enhancement on the second sample EEG signals to obtain sample EEG signals to balance the number of the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase; the second sample EEG signals being the first sample EEG signals in the braking response phase and the first sample EEG signals in the braking perception phase, whichever are smaller; a model training module, configured to input the sample EEG signals into a driving perception classification model to be trained, obtain a sample driving perception stage corresponding to the sample EEG signals, and update the driving perception model to be trained based on the sample driving perception stage corresponding to the sample EEG signals and a label corresponding to the sample EEG signals, until a trained driving perception classification model is obtained; a model application module, configured to input the EEG signal to be tested into the trained driving perception classification model to obtain the driver's driving perception stage corresponding to the EEG signal to be tested, wherein the driving perception stage indicates that the driver is in a danger response stage or a danger perception stage; The driving perception classification model to be trained includes at least: a two-dimensional convolution filter, a depth convolution module, and a separable convolution module; The model training module includes: A first convolution module is configured to input the sample EEG signal into the two-dimensional convolution filter to obtain a plurality of F1 feature maps; the plurality of F1 feature maps correspond to sample EEG signals of different bandpass frequencies; A second convolution module is used to process each of the multiple F1 feature maps through the depth convolution module, and control the number of spatial filters learned by each F1 feature map by depth to obtain a first feature map; a third convolution module, configured to process the first feature map through a separable convolution module to obtain a second feature map; A normalization module is used to perform batch normalization on the second feature map along the feature map dimension to obtain a sample driving perception stage corresponding to the sample EEG signal.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by the processor, the method for recognizing driving perception based on EEG signals as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for recognizing driving perception based on EEG signals as described in any one of claims 1 to 5 is implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method for recognizing driving perception based on EEG signals as described in any one of claims 1 to 5 is implemented.

Images