CN113456078B - Fatigue driving identification method based on PDC dense map propagation - Google Patents
Fatigue driving identification method based on PDC dense map propagation Download PDFInfo
- Publication number
- CN113456078B CN113456078B CN202011287876.0A CN202011287876A CN113456078B CN 113456078 B CN113456078 B CN 113456078B CN 202011287876 A CN202011287876 A CN 202011287876A CN 113456078 B CN113456078 B CN 113456078B
- Authority
- CN
- China
- Prior art keywords
- graph
- electroencephalogram
- signal
- matrix
- band
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 239000011159 matrix material Substances 0.000 claims abstract description 81
- 230000000694 effects Effects 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 19
- 238000010586 diagram Methods 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 11
- 230000001427 coherent effect Effects 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 230000001364 causal effect Effects 0.000 claims description 7
- 230000003993 interaction Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012880 independent component analysis Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 claims description 2
- 238000011176 pooling Methods 0.000 claims description 2
- 206010039203 Road traffic accident Diseases 0.000 abstract 1
- 210000004556 brain Anatomy 0.000 description 10
- 230000008569 process Effects 0.000 description 7
- 230000005611 electricity Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- GNFTZDOKVXKIBK-UHFFFAOYSA-N 3-(2-methoxyethoxy)benzohydrazide Chemical compound COCCOC1=CC=CC(C(=O)NN)=C1 GNFTZDOKVXKIBK-UHFFFAOYSA-N 0.000 description 1
- FGUUSXIOTUKUDN-IBGZPJMESA-N C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 Chemical compound C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 FGUUSXIOTUKUDN-IBGZPJMESA-N 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000013095 identification testing Methods 0.000 description 1
- 230000006996 mental state Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
- A61B5/18—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state for vehicle drivers or machine operators
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Psychiatry (AREA)
- Artificial Intelligence (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Educational Technology (AREA)
- Social Psychology (AREA)
- Child & Adolescent Psychology (AREA)
- Psychology (AREA)
- Mathematical Physics (AREA)
- Hospice & Palliative Care (AREA)
- Developmental Disabilities (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Fuzzy Systems (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention provides a fatigue driving identification method based on PDC dense graph propagation, which comprises the steps of obtaining first electroencephalogram data of a driver at the current time, constructing a first factor connection matrix between first electroencephalograms of each channel by the PDC method, obtaining a feature matrix, extracting space-time features of the electroencephalogram data and information implied between effective connections, constructing a factor connection graph to obtain graph signals, and identifying the graph signals by using trained dense graph propagation to determine whether the driver is in a fatigue driving state. Therefore, the fatigue driving recognition method and the fatigue driving recognition device can improve the recognition accuracy of the fatigue driving, prompt in time in the working environment of the driver and effectively prevent traffic accidents.
Description
Technical Field
The invention belongs to the field of safe driving, and particularly relates to a fatigue driving identification method based on PDC dense map transmission.
Background
At present, the brain electrical signals (electroencephalogram, EEG) can directly and objectively reflect the current mental state of the brain because the brain electrical signals cannot be artificially controlled and forged, and are regarded as 'gold standard' for detecting fatigue driving in a plurality of physiological signal indexes. And different brain electrical data of the driver in a normal state and a fatigue state are analyzed and compared, so that the fatigue driving of the driver can be effectively identified, and further, an effective fatigue early warning indication is made. In recent years, the fatigue driving detection method is mainly researched by collecting data through an electroencephalogram cap, carrying out various feature selections on the collected electroencephalogram data, and then realizing the identification of the fatigue state of a driver through modeling.
In the prior art, the preprocessed electroencephalogram signals are synchronously transmitted to an electroencephalogram signal monitoring device; and extracting 7 characteristic signals from the signals preprocessed by the electroencephalogram signal acquisition device in each second according to the sequence of sampling time, taking the signal values of the 7 characteristic signals as 7 characteristic quantities of the received electroencephalogram signals, then judging the driving fatigue state according to the preset 7 groups of fatigue driving judgment thresholds and calling threshold comparison.
Because the electroencephalogram signal is a spatially discrete and unstable signal, the spatial characteristics of the signal change or the connection of the signal change, so that the accuracy of the result of judging the driving fatigue state in the prior art is reduced.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a fatigue driving identification method based on PDC dense graph propagation. The technical problems to be solved by the invention are realized by the following technical scheme:
The invention provides a fatigue driving identification method based on PDC dense graph propagation, which comprises the following steps:
acquiring first electroencephalogram data of a driver at the current time;
Preprocessing the first electroencephalogram data;
The preprocessed electroencephalogram data is subjected to band-pass filtering to obtain first electroencephalogram signals of all wave bands;
wherein each band includes a plurality of channels;
Based on the first electroencephalogram signals of each channel under each wave band, constructing a first factor connection matrix between the first electroencephalogram signals of each channel by using a partial directional coherent PDC method;
Extracting the characteristics of a first electroencephalogram signal of each channel under each wave band to form a first characteristic matrix;
wherein the features include: power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics;
for each wave band, determining the characteristics in the first characteristic matrix as parameters of the factor connection diagram based on the factor connection diagram corresponding to the first factor connection matrix of the wave band, and obtaining a first diagram signal;
and identifying the first graph signal by using the trained dense graph propagation model, and determining whether the driver is in fatigue driving.
Optionally, the training step of the trained dense graph propagation model includes:
acquiring second electroencephalogram data of the driver in a fatigue state and a normal state;
Preprocessing the second electroencephalogram data;
Band-pass filtering the preprocessed second electroencephalogram data to obtain second electroencephalogram signals of all wave bands;
Based on the second electroencephalogram signals of each channel under each wave band, constructing a second factor connection matrix between the electroencephalogram signals of each channel by using a partial directional coherent PDC method;
Extracting the characteristics of a second electroencephalogram signal of each channel under each wave band to form a second characteristic matrix;
For each wave band, determining the characteristics in the second characteristic matrix as parameters of the factor connection structure model based on the factor connection structure model corresponding to the second factor connection matrix of the wave band, and obtaining a second graph signal;
and iteratively training a preset dense graph propagation model by using the second graph signal to obtain a trained dense graph propagation model.
Optionally, the step of iteratively training a preset dense graph propagation model using the second graph signal to obtain a trained dense graph propagation model includes:
forming the second graph signal into a training set;
Acquiring a test set, wherein the test set is used for determining a graph signal determined by an electroencephalogram signal in a fatigue state or a graph signal determined by an electroencephalogram signal in a normal state;
inputting the training set into a preset dense graph propagation model, and repeatedly adjusting parameters of the dense graph propagation model until the dense graph propagation model after adjusting the parameters recognizes that the accuracy of the test set reaches an accuracy threshold;
And determining the dense graph propagation model with the accuracy reaching the accuracy threshold as a trained dense graph propagation model.
Optionally, the step of preprocessing the first electroencephalogram data includes:
the first electroencephalogram data is preprocessed using artifact subtraction and independent component analysis.
Optionally, the step of obtaining the first electroencephalogram signal of each band by using band-pass filtering on the preprocessed first electroencephalogram data includes:
And carrying out band-pass filtering on the preprocessed electroencephalogram data to obtain first electroencephalogram signals with 4 wave bands.
Optionally, the step of constructing a first causal connection matrix between the electroencephalogram signals of each channel by using a partial directional coherent PDC method based on the first electroencephalogram signals of each channel in each band includes:
for each band, calculating the intensity of the interaction between the first electroencephalogram signals of the channels under the band and determining the direction in which the intensity is calculated;
The intensity of the interaction between the electroencephalogram signals of the channels under each wave band is formed into a factor connection matrix according to the direction of calculating the intensity.
Optionally, the step of extracting characteristics of the electroencephalogram data of each channel in each band to form a characteristic matrix includes:
Extracting power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel under each wave band;
determining the power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel as row elements of a matrix according to the front-to-back sequence;
The row elements are grouped into a feature matrix in the order of the channels.
Optionally, for each band, based on the effect connection graph corresponding to the effect connection matrix of the band, determining the feature in the feature matrix as the parameter of the effect connection graph, and obtaining the first graph signal includes:
constructing a factor connection diagram corresponding to a factor connection matrix of each wave band aiming at each wave band;
The method comprises the steps that a row in an effect connection matrix and a node for representing an effect connection graph are represented, elements in the effect connection matrix are weights of edges between two adjacent nodes in the effect connection matrix, and the effect connection graph comprises a plurality of nodes and edges;
and determining row characteristics in the characteristic matrix as characteristics of nodes based on the corresponding relation between the characteristic matrix rows and the factor connection matrix rows, and obtaining a first graph signal of the factor connection graph.
Optionally, the step of identifying the first graph signal using the trained dense graph propagation model, and determining whether the driver is driving fatigue includes:
identifying the first image signal by using a trained dense image propagation model to obtain a result of whether the first image signal corresponds to the electroencephalogram signal in the fatigue state,
And when the first graph signal corresponds to the electroencephalogram signal in the fatigue state, determining that the driver is in fatigue driving.
Optionally, after the step of determining that the driver is in fatigue driving, the fatigue driving identification method further includes:
and sending an early warning signal to remind the driver.
The invention provides a fatigue driving identification method based on PDC dense graph propagation, which is characterized in that first electroencephalogram data of a driver at the current time are obtained, then a first factor connection matrix between first electroencephalograms of each channel is constructed by the PDC method, then a feature matrix is obtained, information implied between space-time features and effective connections of the electroencephalogram data is extracted, a graph signal is obtained by constructing a factor connection graph, and the graph signal is identified by using trained dense graph propagation, so that whether the driver is in a fatigue driving state or not is determined. The present invention will be described in further detail with reference to the accompanying drawings and examples.
Drawings
FIG. 1 is a flow chart of a fatigue driving identification method based on PDC dense map propagation provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of a connection diagram constructed according to a connection matrix according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a feature imparted to an interconnection graph node according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to specific examples, but embodiments of the present invention are not limited thereto.
Example 1
As shown in fig. 1, the fatigue driving identification method based on PDC dense map propagation provided by the embodiment of the invention includes:
S1, acquiring first electroencephalogram data of a driver at the current time;
It can be understood that the brain electricity data of the person in fatigue and normal state can be acquired by one brain electricity cap acquisition device, and the driver can acquire the first brain electricity data by wearing the brain electricity cap acquisition device.
S2, preprocessing the first electroencephalogram data;
wherein the first electroencephalogram data can be preprocessed using artifact subtraction and independent component analysis.
S3, band-pass filtering is used for the preprocessed electroencephalogram data to obtain first electroencephalogram signals of all wave bands;
the preprocessed electroencephalogram data can be subjected to band-pass filtering to obtain first electroencephalogram signals of 4 wave bands, and each wave band comprises a plurality of channels.
Wherein, 4 wave bands are Theta (4-7 Hz), alpha (8-12 Hz), beta (13-30 Hz), gamma (31-50 Hz) respectively.
S4, constructing a first factor connection matrix between the first electroencephalogram signals of each channel by using a partial directional coherent PDC method based on the first electroencephalogram signals of each channel under each wave band;
s5, extracting the characteristics of the first electroencephalogram signals of each channel under each wave band to form a first characteristic matrix;
wherein the features include: power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics;
s6, for each wave band, determining the characteristics in the first characteristic matrix as parameters of the factor connection diagram based on the factor connection diagram corresponding to the first factor connection matrix of the wave band, and obtaining a first diagram signal;
wherein the parameter is a characteristic of the node.
And S7, identifying the first graph signal by using the trained dense graph propagation model, and determining whether the driver is in fatigue driving.
The invention provides a fatigue driving identification method based on PDC dense graph propagation, which is characterized in that first electroencephalogram data of a driver at the current time are obtained, then a first factor connection matrix between first electroencephalograms of each channel is constructed by the PDC method, then a feature matrix is obtained, information implied between space-time features and effective connections of the electroencephalogram data is extracted, a graph signal is obtained by constructing a factor connection graph, and the graph signal is identified by using trained dense graph propagation, so that whether the driver is in a fatigue driving state or not is determined.
Example two
As an alternative embodiment of the invention, the training step of the trained dense graph propagation model comprises the following steps:
Step a: acquiring second electroencephalogram data of the driver in a fatigue state and a normal state;
Step b: preprocessing the second electroencephalogram data;
step c: band-pass filtering the preprocessed second electroencephalogram data to obtain second electroencephalogram signals of all wave bands;
Step d: based on the second electroencephalogram signals of each channel under each wave band, constructing a second factor connection matrix between the electroencephalogram signals of each channel by using a partial directional coherent PDC method;
Step e: extracting the characteristics of a second electroencephalogram signal of each channel under each wave band to form a second characteristic matrix;
Step f: for each wave band, determining the characteristics in a second characteristic matrix as parameters of the factor connection structure model based on the factor connection structure model corresponding to the second factor connection matrix of the wave band, and obtaining a second graph signal;
step g: and iteratively training a preset dense graph propagation model by using the second graph signal to obtain a trained dense graph propagation model.
It can be understood that in the training process of the model, the electroencephalogram data needs to be processed, and the processing process is the same as the processing process of the electroencephalogram data in the current time, and is the same as the processing process in the first embodiment, and will not be described herein.
Example III
As an optional embodiment of the present invention, the step of iteratively training a preset dense map propagation model using the second map signal to obtain a trained dense map propagation model includes:
Step a: forming a training set by the second graph signals;
step b: acquiring a test set, wherein the graph signals are determined by the electroencephalogram signals in a fatigue state of the test set or the graph signals are determined by the electroencephalogram signals in a normal state;
Step c: inputting the training set into a preset dense graph propagation model, and repeatedly adjusting parameters of the dense graph propagation model until the accuracy of the dense graph propagation model identification test set after the parameters are adjusted reaches an accuracy threshold;
step d: and determining the dense graph propagation model with the accuracy reaching the accuracy threshold as a trained dense graph propagation model.
The preset dense graph propagation model is an improvement on a directed graph in a graph convolution neural network and mainly comprises 5 layers, two convolution layers, two pooling layers and a full connection layer. Unlike traditional graph convolution neural networks, the preset dense graph propagation model does not directly use undirected graphs when input is performed by a convolution layer, but instead introduces two independent connection modes, namely a connection mode from a node to all ancestors of the node, and a connection mode from the node to all offspring. The formula of the dense graph propagation model convolution layer is:
the formula of the preset dense graph propagation model convolution layer is as follows:
In the course of the above-mentioned propagation process, Respectively regularized parent/child adjacency matrices, X ε R N×S representing the feature matrix.
The training set graph signals can be used for carrying out iterative optimization training on a preset dense graph propagation model, and Relu activation functions are used for reducing the dependency relationship among parameters, so that the training accuracy is improved.
Example IV
As an optional embodiment of the present invention, the step of constructing a first causal connection matrix between the electroencephalogram signals of each channel using a partial directional coherent PDC method based on the first electroencephalogram signals of each channel in each band includes:
Step a: for each band, calculating the intensity of the interaction between the first electroencephalogram signals of the channels under the band and determining the direction in which the intensity is calculated;
Step b: the intensity of the interaction between the electroencephalogram signals of the channels under each wave band is formed into a factor connection matrix according to the direction of calculating the intensity.
For example, assume that the channel under each band is 3, which is 1, 2, and 3, respectively, and the electroencephalogram signal of the channel 1 is: {1.23,1.34,4.12, … }; the electroencephalogram signal of the channel 2 is: {5.2,1.76,0.2, … }; the electroencephalogram signal of the channel 3 is: {6.4,3.65,3.76, … }, since the effective connection matrix is:
the intensity of the 1- >2 channels is 0.5, the intensity of the 2- >1 channels is 0.7, the directions are different between the intensity of the electroencephalogram signals among the 1- >2 channels and the intensity of the 2- >1 channels, and the signals are directional.
Example five
As an optional embodiment of the present invention, the step of extracting characteristics of brain electrical data of each channel in each band to form a characteristic matrix includes:
Step a: extracting power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel under each wave band;
Step b: determining the power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel as row elements of a matrix according to the front-to-back sequence;
step c: the row elements are grouped into a feature matrix in the order of the channels.
Wherein, the number of channels is N, the feature dimension is D dimension, and then the features of all channels form an N x D feature matrix.
For example, assuming that the number of channels is 3, the four features described above are extracted, and the resulting feature matrix is 3×4, as follows:
Modeling the extracted brain electrical characteristics of each channel into graph signals based on the factor connection mode structure; each electroencephalogram channel corresponds to a node in the graph, four features extracted by each channel serve as features of the node, and the causal connection between two different nodes corresponds to one edge in the graph.
Example six
As an optional embodiment of the present invention, for each band, the step of determining, based on the effect connection graph corresponding to the effect connection matrix of the band, the feature in the feature matrix as a parameter of the effect connection graph, and obtaining the first graph signal includes:
Step a: constructing a factor connection diagram corresponding to a factor connection matrix of each wave band aiming at each wave band;
The method comprises the steps that a row in an effect connection matrix and a node for representing an effect connection graph are displayed, elements in the effect connection matrix are weights of edges between two adjacent nodes in the effect connection matrix, and the effect connection graph comprises a plurality of nodes and edges;
step b: and determining row characteristics in the characteristic matrix as characteristics of nodes based on the corresponding relation between the characteristic matrix rows and the factor connection matrix rows, and obtaining a first graph signal of the factor connection graph.
The efficient connection graph consists of nodes and edges, g= (V, E), V is the set of nodes, and E is the set of weights for the edges. Each channel represents a node in the graph, eij represents the weight from node i to node j, and the weight from node 1 to node 2 is 0.5 as follows, and the relationship between the effective connection matrix and the effective connection graph is as follows:
As shown in fig. 2, fig. 2 is a graph of the effect connection constructed from the effect connection matrix. The graph signal is given as an adjacency matrix A of any graph, and given that the node number is N, any signal with the length of N can be used as the graph signal of the factor connection graph, and the value of each position represents the value of the corresponding node. The number of the corresponding nodes of the electroencephalogram signals is 3, the signals are four features extracted from each channel, the length is 4, the value of the position of the node 1 is {1.2,0.5,2.3,2.2}, and the feature matrix is:
Thus, node 1 corresponds to lane 1, node 2 corresponds to lane 2, and node 3 corresponds to lane 3. As shown in fig. 3, the structure based on the cause-effect connection mode is to first establish a graph-effect connection graph, and then assign a value to each graph node on the basis of the graph, namely, the characteristic of the electroencephalogram signal under the channel, and this process is called a graph signal obtaining process.
Example seven
As an alternative embodiment of the present invention, the step of identifying the first graph signal using the trained dense graph propagation model, and determining whether the driver is driving fatigue includes:
Step a: identifying the first image signal by using the trained dense image propagation model to obtain a result of whether the first image signal corresponds to the electroencephalogram signal in the fatigue state;
Step b: and when the first graph signal corresponds to the electroencephalogram signal in the fatigue state, determining that the driver is in fatigue driving.
Example eight
As an optional embodiment of the present invention, after the step of determining that the driver is in the fatigue driving, the fatigue driving identification method further includes:
and sending an early warning signal to remind the driver.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
In the present invention, unless expressly stated or limited otherwise, a first feature "above" or "below" a second feature may include both the first and second features being in direct contact, as well as the first and second features not being in direct contact but being in contact with each other through additional features therebetween. Moreover, a first feature being "above," "over" and "on" a second feature includes the first feature being directly above and obliquely above the second feature, or simply indicating that the first feature is higher in level than the second feature. The first feature being "under", "below" and "beneath" the second feature includes the first feature being directly under and obliquely below the second feature, or simply means that the first feature is less level than the second feature.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Further, one skilled in the art can engage and combine the different embodiments or examples described in this specification.
Although the application is described herein in connection with various embodiments, other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed application, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.
The foregoing is a further detailed description of the invention in connection with the preferred embodiments, and it is not intended that the invention be limited to the specific embodiments described. It will be apparent to those skilled in the art that several simple deductions or substitutions may be made without departing from the spirit of the invention, and these should be considered to be within the scope of the invention.
Claims (8)
1. The fatigue driving identification method based on PDC dense map propagation is characterized by comprising the following steps of:
acquiring first electroencephalogram data of a driver at the current time;
Preprocessing the first electroencephalogram data;
The preprocessed electroencephalogram data is subjected to band-pass filtering to obtain first electroencephalogram signals of all wave bands;
wherein each band includes a plurality of channels;
Based on the first electroencephalogram signals of each channel under each wave band, constructing a first factor connection matrix between the first electroencephalogram signals of each channel by using a partial directional coherent PDC method;
Extracting the characteristics of a first electroencephalogram signal of each channel under each wave band to form a first characteristic matrix;
wherein the features include: power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics;
for each wave band, determining the characteristics in the first characteristic matrix as parameters of the factor connection diagram based on the factor connection diagram corresponding to the first factor connection matrix of the wave band, and obtaining a first diagram signal;
identifying the first graph signal by using the trained dense graph propagation model, and determining whether a driver is in fatigue driving or not;
The dense graph propagation model is an improvement on a directed graph in a graph convolution neural network and comprises 5 layers, two convolution layers, two pooling layers and a full connection layer; when the convolution layer inputs, two independent connection modes are introduced, namely a connection mode from a node to all ancestor of the node, and a connection mode from the node to all offspring;
the step of extracting the characteristics of the electroencephalogram data of each channel under each wave band to form a characteristic matrix comprises the following steps:
Extracting power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel under each wave band;
determining the power spectral density, differential entropy, differential asymmetry and rational asymmetry characteristics of the electroencephalogram data of each channel as row elements of a matrix according to the front-to-back sequence;
the line elements form a feature matrix according to the order of the channels;
The step of determining, for each band, the feature in the feature matrix as the parameter of the causal connection graph based on the causal connection graph corresponding to the causal connection matrix of the band, and obtaining the first graph signal includes:
constructing a factor connection diagram corresponding to a factor connection matrix of each wave band aiming at each wave band;
The method comprises the steps that a row in an effect connection matrix and a node for representing an effect connection graph are represented, elements in the effect connection matrix are weights of edges between two adjacent nodes in the effect connection matrix, and the effect connection graph comprises a plurality of nodes and edges;
and determining row characteristics in the characteristic matrix as characteristics of nodes based on the corresponding relation between the characteristic matrix rows and the factor connection matrix rows, and obtaining a first graph signal of the factor connection graph.
2. The fatigue driving identification method according to claim 1, wherein the training step of the trained dense map propagation model includes:
acquiring second electroencephalogram data of the driver in a fatigue state and a normal state;
Preprocessing the second electroencephalogram data;
Band-pass filtering the preprocessed second electroencephalogram data to obtain second electroencephalogram signals of all wave bands;
Based on the second electroencephalogram signals of each channel under each wave band, constructing a second factor connection matrix between the electroencephalogram signals of each channel by using a partial directional coherent PDC method;
Extracting the characteristics of a second electroencephalogram signal of each channel under each wave band to form a second characteristic matrix;
For each wave band, determining the characteristics in the second characteristic matrix as parameters of the factor connection structure model based on the factor connection structure model corresponding to the second factor connection matrix of the wave band, and obtaining a second graph signal;
and iteratively training a preset dense graph propagation model by using the second graph signal to obtain a trained dense graph propagation model.
3. The fatigue driving recognition method according to claim 1, wherein the step of iteratively training a preset dense map propagation model using the second map signal to obtain a trained dense map propagation model comprises:
forming the second graph signal into a training set;
Acquiring a test set, wherein the test set is used for determining a graph signal determined by an electroencephalogram signal in a fatigue state or a graph signal determined by an electroencephalogram signal in a normal state;
inputting the training set into a preset dense graph propagation model, and repeatedly adjusting parameters of the dense graph propagation model until the dense graph propagation model after adjusting the parameters recognizes that the accuracy of the test set reaches an accuracy threshold;
And determining the dense graph propagation model with the accuracy reaching the accuracy threshold as a trained dense graph propagation model.
4. The fatigue driving identification method according to claim 1, wherein the step of preprocessing the first electroencephalogram data includes:
the first electroencephalogram data is preprocessed using artifact subtraction and independent component analysis.
5. The fatigue driving recognition method according to claim 1, wherein the step of obtaining the first electroencephalogram signal of each band by band-pass filtering the preprocessed first electroencephalogram data includes:
And carrying out band-pass filtering on the preprocessed electroencephalogram data to obtain first electroencephalogram signals with 4 wave bands.
6. The fatigue driving recognition method according to claim 1, wherein the step of constructing a first causal connection matrix between the electroencephalogram signals of the channels using a partial directional coherent PDC method based on the first electroencephalogram signals of the channels in each band comprises:
for each band, calculating the intensity of the interaction between the first electroencephalogram signals of the channels under the band and determining the direction in which the intensity is calculated;
The intensity of the interaction between the electroencephalogram signals of the channels under each wave band is formed into a factor connection matrix according to the direction of calculating the intensity.
7. The method of claim 1, wherein the step of identifying the first map signal using the trained dense map propagation model, and determining whether the driver is driving fatigue comprises:
identifying the first image signal by using a trained dense image propagation model to obtain a result of whether the first image signal corresponds to the electroencephalogram signal in the fatigue state,
And when the first graph signal corresponds to the electroencephalogram signal in the fatigue state, determining that the driver is in fatigue driving.
8. The fatigue driving recognition method according to claim 1, wherein after the step of determining that the driver is in fatigue driving, the fatigue driving recognition method further comprises:
and sending an early warning signal to remind the driver.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011287876.0A CN113456078B (en) | 2020-11-17 | 2020-11-17 | Fatigue driving identification method based on PDC dense map propagation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011287876.0A CN113456078B (en) | 2020-11-17 | 2020-11-17 | Fatigue driving identification method based on PDC dense map propagation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113456078A CN113456078A (en) | 2021-10-01 |
| CN113456078B true CN113456078B (en) | 2024-06-11 |
Family
ID=77868246
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011287876.0A Active CN113456078B (en) | 2020-11-17 | 2020-11-17 | Fatigue driving identification method based on PDC dense map propagation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113456078B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115620484A (en) * | 2022-09-30 | 2023-01-17 | 重庆长安汽车股份有限公司 | Fatigue driving alarm method and device, vehicle-mounted unit system, vehicle and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104473636A (en) * | 2014-12-30 | 2015-04-01 | 天津大学 | Brain fatigue network analysis method based on partial orientation coherence |
| CN109770924A (en) * | 2019-01-24 | 2019-05-21 | 五邑大学 | A kind of tired classification method based on Hadamard product building brain function network and Method Using Relevance Vector Machine |
| CN110399857A (en) * | 2019-08-01 | 2019-11-01 | 西安邮电大学 | A kind of brain electricity emotion identification method based on figure convolutional neural networks |
| CN110969108A (en) * | 2019-11-25 | 2020-04-07 | 杭州电子科技大学 | Limb action recognition method based on autonomic motor imagery electroencephalogram |
| CN111544017A (en) * | 2020-05-25 | 2020-08-18 | 五邑大学 | GPDC graph convolution neural network-based fatigue detection method and device and storage medium |
-
2020
- 2020-11-17 CN CN202011287876.0A patent/CN113456078B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104473636A (en) * | 2014-12-30 | 2015-04-01 | 天津大学 | Brain fatigue network analysis method based on partial orientation coherence |
| CN109770924A (en) * | 2019-01-24 | 2019-05-21 | 五邑大学 | A kind of tired classification method based on Hadamard product building brain function network and Method Using Relevance Vector Machine |
| CN110399857A (en) * | 2019-08-01 | 2019-11-01 | 西安邮电大学 | A kind of brain electricity emotion identification method based on figure convolutional neural networks |
| CN110969108A (en) * | 2019-11-25 | 2020-04-07 | 杭州电子科技大学 | Limb action recognition method based on autonomic motor imagery electroencephalogram |
| CN111544017A (en) * | 2020-05-25 | 2020-08-18 | 五邑大学 | GPDC graph convolution neural network-based fatigue detection method and device and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 一种与N-Back诱发脑力疲劳相关的特异性脑电网络结构;孟桂芳 等;中国生物医学工程学报;20170430;全文 * |
| 体疲劳对脑疲劳影响的脑电信息分析与处理;张春翠;中国优秀硕士学位论文全文数据库医药卫生科技辑;20170315;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113456078A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111310786B (en) | Traffic detector abnormality diagnosis method and device based on random forest classifier | |
| CN107506692A (en) | A kind of dense population based on deep learning counts and personnel's distribution estimation method | |
| CN104720796A (en) | Automatic detecting system and method for epileptic attack time period | |
| CN113842118B (en) | Epileptic seizure real-time detection monitoring system for epileptic video electroencephalogram examination | |
| KR102077605B1 (en) | Method and apparatus for an automatic artifact removal of EEG based on a deep learning algorithm | |
| CN108903936A (en) | The Intelligent mining helmet control method merged based on human body information and environmental information | |
| CN104008633B (en) | A kind of facility spinach diseases method for early warning and system | |
| CN106934199A (en) | A kind of psychological pressure appraisal procedure and device | |
| CN108599863B (en) | Overhead transmission line wind dance monitoring and early warning system and method based on shallow CNN classification model | |
| CN103006211A (en) | Map mapping device based on brain electrical activity network analysis | |
| US20220067584A1 (en) | Model generation apparatus, model generation method, computer-readable storage medium storing a model generation program, model generation system, inspection system, and monitoring system | |
| CN116313132B (en) | Medical management system for chronic diseases | |
| CN112294337A (en) | Epilepsy detection system based on linear graph convolution neural network | |
| CN114743678B (en) | Intelligent bracelet physiological index anomaly analysis method and system based on improved GDN algorithm | |
| CN113069117A (en) | Electroencephalogram emotion recognition method and system based on time convolution neural network | |
| CN113456078B (en) | Fatigue driving identification method based on PDC dense map propagation | |
| CN117671396A (en) | Intelligent monitoring and early warning system and method for construction progress | |
| KR20190073330A (en) | Method and apparatus for an automatic artifact removal of EEG based on a deep leaning algorithm | |
| CN104794505A (en) | Multichannel electroencephalogram data fusion and dimension descending method | |
| CN112613749B (en) | Cross-border hidden high-risk factor risk intelligent analysis system | |
| CN111387975A (en) | Electroencephalogram signal identification method based on machine learning | |
| CN116295847A (en) | Power equipment fault diagnosis system and diagnosis method | |
| KR20200099811A (en) | Apparatus and method for providing information on epilepsy symptoms based on resting-state EEG | |
| CN118332519A (en) | Distributed photovoltaic anomaly detection method and device based on split-level cyclic neural network | |
| CN118038548A (en) | Abnormal behavior detection method, device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |