CN113143275B - Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner - Google Patents

Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner Download PDF

Info

Publication number
CN113143275B
CN113143275B CN202110317792.5A CN202110317792A CN113143275B CN 113143275 B CN113143275 B CN 113143275B CN 202110317792 A CN202110317792 A CN 202110317792A CN 113143275 B CN113143275 B CN 113143275B
Authority
CN
China
Prior art keywords
formula
sample
electroencephalogram
equation
theta
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
Application number
CN202110317792.5A
Other languages
Chinese (zh)
Other versions
CN113143275A (en
Inventor
彭勇
李幸
张怿恺
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN202110317792.5A priority Critical patent/CN113143275B/en
Publication of CN113143275A publication Critical patent/CN113143275A/en
Application granted granted Critical
Publication of CN113143275B publication Critical patent/CN113143275B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • 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
    • A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
    • 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

Abstract

The invention provides an electroencephalogram fatigue detection method for joint quantitative evaluation of samples and characteristic quality. The invention comprises the following steps: 1. and a plurality of testees respectively acquire electroencephalogram data under the simulated driving system. 2. And (3) preprocessing and extracting the characteristics of all the electroencephalogram data obtained in the step (1). 3. And establishing a machine learning model to realize electroencephalogram fatigue detection of sample and characteristic quality combined quantitative evaluation. 4. And solving a description factor v for measuring the quality of the sample and a description factor theta for the characteristics. 5. And carrying out fatigue regression prediction on the new electroencephalogram data of the testee. According to the method, after v and theta are embedded into a least square model, the obtained weight description factors for measuring the quality and the characteristics of the sample provide an effective tool for performing electroencephalogram data sample selection and characteristic selection, higher weight is given to the sample and the characteristics with better quality, and the fatigue condition of the tested person can be accurately obtained according to the electroencephalogram data.

Description

Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner
Technical Field
The invention belongs to the technical field of electroencephalogram signal processing, and particularly relates to an electroencephalogram fatigue detection method for joint quantitative evaluation of samples and characteristic quality.
Background
With the increasing development of the traffic industry of China, the research of more practical and objective driving fatigue detection has important significance for improving the active traffic safety. The comprehensive domestic and foreign fatigue detection method mainly comprises subjective evaluation and objective detection. The subjective evaluation method is mainly used for judging whether the patient is in a fatigue state or not by recording subjective questionnaires such as a Pearson fatigue scale, a Stanford sleep scale and the like. The field of objective detection is mainly divided into: detection based on vehicle behavior characteristics, detection based on driver characteristic behavior characteristics, detection based on physiological electrical characteristics. The electroencephalogram signal in the objective detection method is used as the representation of the central nervous signal activity, the method has the characteristic of high detection accuracy, and the method is a gold standard for fatigue judgment. However, the quality of the acquired electroencephalogram data is lack of reliability due to the influence of cross-time acquisition and the position of an electrode cap in the current electroencephalogram acquisition process.
The electroencephalogram signals serve as unsteady-state signals, the samples contain more noise, if description factors of sample quality and sample characteristics can be described for each electroencephalogram data in the learning process, the electroencephalogram data quality can be distinguished, and the characteristics beneficial to model training are selected, so that the performance of a machine learning model is improved, and the model has better robustness.
Disclosure of Invention
The invention aims to provide an electroencephalogram fatigue detection method for joint quantitative evaluation of a sample and characteristic quality. By the method, a description factor v for measuring the quality of the sample and a description factor theta for measuring the characteristics of the sample can be obtained, a least square model is embedded, a quantization factor v is obtained through calculation to depict the quality of the sample, and the description factor theta for measuring the characteristics of the sample is obtained through self-learning, so that the negative influence of a large-noise electroencephalogram sample on the robustness of the model is avoided, and the obtained description factor is used for selecting the sample to obtain higher fatigue detection precision.
The method comprises the following specific steps:
step 1, carrying out electroencephalogram data acquisition on a plurality of testees in the process of gradual fatigue.
And 2, preprocessing and extracting characteristics of all the electroencephalogram data obtained in the step 1. Each processed group of data is taken as a sample matrix X. Each sample matrix X corresponds to a label vector y; the tag vector y corresponds to the fatigue level of the subject.
And 3, establishing a machine learning model for electroencephalogram fatigue detection.
3-1, establishing an embedded describing factor v and a theta objective function as shown in the formula (1).
Figure BDA0002990580690000021
In the formula (1), the description factor theta is determined according to the description matrix theta, and the specific relation is
Figure BDA0002990580690000022
ΘjjIs the jth diagonal element of the description matrix Θ; thetajIs the jth element describing the factor θ;
Figure BDA0002990580690000023
w and b are weight and deviation respectively; Θ is a description matrix; x is the number ofiIs the ith column element of the sample matrix X; y isiIs the ith element of label y; v. ofiDescribing the quality of each sample in the sample matrix for the ith element of the description factor v;
Figure BDA0002990580690000024
represents a square calculation of the 2 norm; gamma represents a regularization term parameter; f (λ, v) represents a regular term function.
3-2, establishing an expression of a regular term function f (lambda, v) as shown in the formula (2):
Figure BDA0002990580690000025
in the formula (2), n is the number of samples in the sample matrix; λ is a parameter in the regularizing term function f (λ, v).
3-3. through fixing v,
Figure BDA0002990580690000026
b, writing the formula (1) as shown in the formula (3):
Figure BDA0002990580690000027
the optimal solution of θ obtained by the lagrange multiplier method is shown in formula (4):
Figure BDA0002990580690000028
in the formula (4), θiTo describe the ith element of the factor θ, the importance of each feature is described.
Figure BDA00029905806900000212
Is a variable of
Figure BDA00029905806900000210
The ith element of (1).
The objective function formula (1) is then further rewritten into formula (5).
Figure BDA0002990580690000029
And 4, obtaining the updating rules of all variables by fixing other variables and reserving only one variable according to the target function shown in the formula (5), and further obtaining the variable in the formula (5)
Figure BDA00029905806900000211
The factor v, the deviation b, is described.
Step 5, carrying out electroencephalogram data acquisition on the testee; inputting the obtained electroencephalogram data test set serving as a sample matrix into a computer for determining v, theta,
Figure BDA00029905806900000311
And b, acquiring a predicted value label corresponding to the electroencephalogram data in the target function formula (5). And the predicted value label is the detected fatigue degree of the testee.
Preferably, in step 2, the subject is gradually fatigued by continuing the simulated driving on a straight road with the simulated driving environment.
Preferably, the label vector y in step 2 is determined by the duration of the eye closure of the subject during the electroencephalogram data acquisition process. The label y is represented by a value between 0 and 1.
Preferably, v is determined in step 4,
Figure BDA00029905806900000314
The specific process of b is as follows;
4-1, fixing the steel wire by a fixing device b,
Figure BDA00029905806900000312
To update v such that equation (5) is written as shown in equation (6);
Figure BDA0002990580690000031
in the formula (6)
Figure BDA0002990580690000032
Deriving v in the formula (6) to obtain an update rule of v;
Figure BDA0002990580690000033
wherein liA loss function for a machine learning model expressed as
Figure BDA0002990580690000034
4-2. through fixing
Figure BDA00029905806900000313
v updates b so that equation (5) is written as shown in equation (8);
Figure BDA0002990580690000035
let diagonal matrix
Figure BDA0002990580690000036
So that equation (8) is written as shown in equation (9):
Figure BDA0002990580690000037
taking b in equation (9) as a derivative and letting the derivative be 0, the update rule for b is given by equation (10):
b＝(UTU)-1(UTU)(y-XTw) (10)
4-3. update by fixing v, b
Figure BDA00029905806900000315
Such that equation (10) is written as shown in equation (11);
Figure BDA0002990580690000038
let diagonal matrix
Figure BDA0002990580690000039
So that equation (11) is written as shown in equation (12):
Figure BDA00029905806900000310
to the formula (12)
Figure BDA0002990580690000042
Derivative and let the derivative be 0, get
Figure BDA0002990580690000043
The update rule of (2) is equation (13):
Figure BDA0002990580690000044
preferably, the pretreatment in step 2 is carried out as follows:
2-1, down-sampling the electroencephalogram data to 200Hz, and performing band-pass filtering on the electroencephalogram data to a range of 1-50 Hz; according to the 5-frequency band method, the method is divided into five frequency bands of Delta, Theta, Alpha, Beta and Gamma
2-2, respectively carrying out short-time Fourier transform with 4 seconds of time window and no overlap on the electroencephalogram data of the 5 frequency bands, and extracting differential entropy characteristics h (X) as shown in a formula (14):
h(X)＝-∫xf(x)ln(f(x))dx (14)
in the formula (14), X is an input sample matrix, and c is an element in the input sample matrix; (x) is a probability density function;
the updated differential entropy characteristic h (X) is shown as a formula (15);
Figure BDA0002990580690000041
in the formula (15), σ is a standard deviation of the probability density function; μ is the expectation of the probability density function.
Preferably, 17 leads are adopted for the electroencephalogram data acquisition, and 5 frequency bands are selected; the 5 frequency bands are respectively 1-4Hz, 4-8Hz, 8-14Hz, 14-31Hz and 31-50 Hz.
The invention has the beneficial effects that:
1. the description factor for measuring the quality and the characteristics of the sample, which is obtained by the invention, provides an effective tool for selecting the sample of the unsteady electroencephalogram data, and the sample with better quality is given higher weight through the mathematical model, so that the fatigue condition of the tested person can be accurately obtained according to the electroencephalogram data.
2. Aiming at the characteristic of more noise of electroencephalogram signals, the invention sets a description factor for describing sample quality and sample characteristics, wherein v is obtained by measuring and calculating a loss function, and when the loss function is larger, a smaller upsilon can be obtained by a formula (7)iTherefore, the proportion of the noise sample points in the loss function is small, and the denoising effect is achieved; the value of theta is obtained by self-learning, and the theta can be known from the formula (4)iIs only equal to
Figure BDA0002990580690000045
Is concerned with when
Figure BDA0002990580690000046
When the value in (5) is larger, θ can be found from the formula (4)iThe larger the value of (A) is, the
Figure BDA0002990580690000047
The value of (A) is shown in formula (1), and the process explains the reason why theta is explained from the point of view of mathematical theoryiThe sample matrix can be subjected to feature selection, so that the quality and the features of different electroencephalogram data are distinguished, and the identification effect of the machine learning model on the electroencephalogram signals is improved.
3. The electroencephalogram data acquisition process is achieved by a plurality of electrode caps, the sample data is influenced by experiment time and lead positions, and each lead represents a characteristic dimension. According to the method, the lead position which is more favorable for model training can be obtained by performing feature selection on the sample, namely, the sample is subjected to feature selection, so that the robustness and the accuracy of the model are improved in the training process.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
fig. 2 is a schematic diagram of an embedding process of two description factors respectively corresponding to the sample quality and the feature importance degree in the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
The invention solves the important problem of important characteristic excavation of the electroencephalogram signal in fatigue detection based on the following starting points: in fatigue detection, electroencephalogram signals are used as unsteady-state signals, samples contain more noise, and if the quality of each sample can be characterized and the characteristic dimension of each sample can be selected in the learning process, the samples and the characteristics which are beneficial to model training are selected, and a model with better robustness can be obtained. Therefore, the samples and the characteristics with good quality can be selected for learning, and the method has important significance for improving the accuracy of fatigue detection.
As shown in fig. 1 and 2, a electroencephalogram fatigue detection method for joint quantitative evaluation of a sample and characteristic quality specifically comprises the following steps:
step 1, acquiring electroencephalogram data of the induced fatigue state of a subject by using a simulated driving system (in the embodiment, a simulated driving scene containing obvious induced fatigue of the subject is used).
The method comprises the steps of carrying out M times of electroencephalogram data acquisition on N subjects in the same simulated driving environment to obtain N.M groups of electroencephalogram data, wherein the data volume of each group of data is d x N, d is the dimensionality of each group of data, and N is the number of electroencephalogram data samples which are acquired at a single time and are related to time. The set of data includes electroencephalogram data for a plurality of time instants obtained in one acquisition. Each set of data is taken as a sample matrix X. Each sample matrix X corresponds to a label y; the label y corresponds to the fatigue degree of the testee, and is determined by the percentage of the time length of the closed eyes of the testee in the electroencephalogram data acquisition process to the total acquisition time length. Longer duration of eye closure indicates more fatigue in the subject.
Unifying the corresponding tested task scenes of the tested subjects in the testing task, wherein each tested subject completes the testing of all the tested task scenes. In this embodiment, the task scene to be tested is a straight and monotonous road, and the experiment time is performed after lunch, so that the subject is more fatigued. The content of these different batches of experiments is the same, but their status at different dates may be different; this reflects that the essential features of the brain electricity will not change greatly with the change of time or experimental scenes. These differences in time, or in the scene during the experiment, are the differentiation components.
And 2, preprocessing and extracting characteristics of all the electroencephalogram data obtained in the step 1. The method is carried out on the basis of 17-lead and 5-frequency bands (Delta (1-4Hz), Theta (4-8Hz), Alpha (8-14Hz), Beta (14-31Hz) and Gamma (31-50Hz)) and by extracting differential entropy characteristics. In practical application, the number of leads depends on the electroencephalogram cap worn by a subject during data acquisition; the division of frequency bands also follows a physiologically meaningful 5-band division; the most common features of electroencephalographic signals are power spectral density and differential entropy. The electroencephalogram signal of a human being is very weak, which means that the electroencephalogram signal is easy to interfere, and the acquired result is difficult to directly carry out experiments, so that the requirements on the preprocessing of the electroencephalogram signal are provided:
the pretreatment process is as follows:
2-1, down-sampling the electroencephalogram data to 200Hz, and performing band-pass filtering on the electroencephalogram data to a range of 1-50 Hz. According to the 5-frequency band method, the method is divided into five frequency bands of Delta, Theta, Alpha, Beta and Gamma
And 2-2, taking the electroencephalogram data of the 5 frequency bands as sample matrixes, respectively carrying out short-time Fourier transform with 4 seconds of time window and no overlap, and extracting differential entropy characteristics. The differential entropy signature h (x) is defined as:
h(X)＝-∫xf(x)ln(f(x))dx (14)
in the formula (14), X is an input sample matrix (i.e. electroencephalogram data of a certain frequency band), and X is an element in the input sample matrix; f (x) is a probability density function. For a sample matrix X following a gaussian distribution, its differential entropy characteristic h (X) can be calculated as shown in equation (15):
Figure BDA0002990580690000061
in the formula (15), σ is a standard deviation of the probability density function; μ is the expectation of the probability density function.
It can be seen that the differential entropy signature is essentially a logarithmic form of the power spectral density signature, i.e.
Figure BDA0002990580690000064
The preprocessing of the electroencephalogram signals aims to improve the signal-to-noise ratio, so that the preprocessing effect of data is improved, and interference is reduced.
And 3, establishing a machine learning model to realize electroencephalogram fatigue detection based on sample and characteristic quality combined quantitative evaluation. Describing factors v and theta for measuring the quality and the characteristics of the sample have nonnegative and normalized characteristics; the ith element v describing the factor viTo characterize the mass of the ith sample; ith sample element theta describing factor thetaiTo characterize the importance of the i-th dimension of the sample. Then v isiAnd thetaiAnd embedding the electroencephalogram fatigue detection model into a least square learning model to obtain the electroencephalogram fatigue detection model of the sample and characteristic quality combined quantitative evaluation as shown in the formula (16).
3-1, embedding the description factors v and theta into a least square model to obtain a target function of the electroencephalogram fatigue detection method for the combined quantitative evaluation of the sample and the characteristic quality, wherein the target function is as shown in a formula (5):
Figure BDA0002990580690000062
in the formula (16), the compound represented by the formula,
Figure BDA0002990580690000063
is a loss function of a least squares model, wherein,
Figure BDA0002990580690000065
and
Figure BDA0002990580690000066
respectively weight and deviation in the least square learning model;
Figure BDA0002990580690000078
a description matrix representing characteristics of the sample; x is the number ofiIs the ith column element of the sample matrix X; y isiIs the ith element of label y;
Figure BDA0002990580690000079
a matrix of samples is represented that is,
Figure BDA00029905806900000710
a representation tag (i.e., a sample tag value); upsilon isiIn order to correspond to a descriptive factor of the sample quality,
Figure BDA0002990580690000076
represents a square calculation of the 2 norm; γ represents a regularization term parameter for balancing the weights; f (λ, v) represents a canonical term function describing the factor v.
The target function formula (5) is obtained by adding a regular term function on the basis of the initial target function formula (17);
Figure BDA0002990580690000071
as can be seen from the formula (17), when viWhen all the target functions are 0, the target function formula (17) is minimum and is not in accordance with the requirement, so a regular term is added to the target function, and the target function is updated to be formula (5).
3-2, establishing an expression of a regular term function f (lambda, v) as shown in the formula (2):
Figure BDA0002990580690000072
in the formula (2), n is the number of samples in a sample matrix; upsilon isiRepresenting any descriptive factor, upsilon, that measures the quality of the sampleiIs calculated, and v is calculated as the better the quality of the sample isiThe closer to 1 the magnitude of (c) is, the worse the quality of the samples, the closer to 0, each sample having a describing factor vi(ii) a λ is a parameter in the regularizing term function f (λ, v).
3-3. order
Figure BDA0002990580690000077
Equation (16) can be written as shown in equation (1):
Figure BDA0002990580690000073
determining a description factor theta according to the description matrix theta, wherein the specific relation is
Figure BDA00029905806900000711
ΘjjIs the jth diagonal element of the description matrix Θ; thetajTo describe the jth element of the factor theta.
By fixing v, w, b, equation (1) can be written as shown in equation (3):
Figure BDA0002990580690000074
the optimal solution of θ obtained by the lagrange multiplier method is shown in formula (4):
Figure BDA0002990580690000075
equation (1) can be written as
Figure BDA0002990580690000081
And 4, obtaining the updating rules of all variables by a method of fixing other variables and only reserving one variable according to the target function shown in the formula (16), and further obtaining a description factor v for measuring the quality of the sample and the weight w and the deviation b of the least square model. Find v,
Figure BDA00029905806900000810
The specific process of b is as follows;
4-1, fixing the steel wire by a fixing device b,
Figure BDA00029905806900000811
To update v, equation (5) may be written as shown in equation (6);
Figure BDA0002990580690000082
in the formula (6)
Figure BDA0002990580690000083
The derivative of v in the formula (5) can obtain the update rule of v as;
Figure BDA0002990580690000084
wherein liA loss function for a machine learning model expressed as
Figure BDA00029905806900000812
4-2. through fixing
Figure BDA00029905806900000813
v to update b, equation (5) may be written as shown in equation (8);
Figure BDA0002990580690000085
let diagonal matrix
Figure BDA0002990580690000086
Equation (8) can be written as shown in equation (9):
Figure BDA0002990580690000087
taking b in equation (9) as a derivative and letting the derivative be 0, we can obtain the update rule of b as equation (10):
b＝(UTU)-1(UTU)(y-XTw) (10)
4-3. update by fixing v, b
Figure BDA00029905806900000817
Equation (5) can be written as shown in equation (11);
Figure BDA0002990580690000088
let diagonal matrix
Figure BDA00029905806900000814
Equation (11) can be written as shown in equation (12):
Figure BDA0002990580690000089
to the formula (12)
Figure BDA00029905806900000815
Taking the derivative and let it be 0, we can get
Figure BDA00029905806900000816
The update rule of (2) is equation (13):
Figure BDA0002990580690000091
step 5, carrying out electroencephalogram data acquisition on the testee; inputting the obtained electroencephalogram data test set as a sample matrix to the test set in which v, theta, and the like have been determined in the fourth step,
Figure BDA0002990580690000092
And b, acquiring a predicted value label y corresponding to the electroencephalogram data in the target function formula (5). And the predicted value label is the detected fatigue degree of the testee. Because the description factors v and theta distinguish the importance of each sample and feature in the sample matrix, the accuracy of fatigue detection can be effectively improved.

Claims (6)

1. A electroencephalogram fatigue detection method based on sample and characteristic quality combined quantitative evaluation is characterized by comprising the following steps: step 1, carrying out electroencephalogram data acquisition on a plurality of testees in a gradual fatigue process;
step 2, preprocessing and feature extraction are carried out on all the electroencephalogram data obtained in the step 1; each processed group of data is used as a sample matrix X; each sample matrix X corresponds to a label vector y; the label vector y corresponds to the fatigue degree of the subject;
step 3, establishing a machine learning model for electroencephalogram fatigue detection;
3-1, establishing an embedded description factor v and a theta objective function as shown in a formula (1);
Figure FDA0003555673280000011
in the formula (1), the description factor theta is determined according to the description matrix theta, and the specific relation is
Figure FDA0003555673280000012
ΘjjIs the jth diagonal element of the description matrix Θ; thetajIs the jth element describing the factor θ;
Figure FDA0003555673280000013
and
Figure FDA0003555673280000014
respectively weight and deviation in the least square learning model; Θ is a description matrix; x is the number ofiIs the ith column element of the sample matrix X; y isiIs the ith element of label y; v. ofiDescribing the quality of each sample in the sample matrix for the ith element of the description factor v;
Figure FDA0003555673280000015
represents a square calculation of the 2 norm; gamma represents a regularization term parameter; f (λ, v) represents a regular term function; describing that the factors v and theta have non-negative and normalized characteristics;
3-2, establishing an expression of a regular term function f (lambda, v) as shown in the formula (2):
Figure FDA0003555673280000016
in the formula (2), n is the number of samples in the sample matrix; λ is a parameter in the regularization term function f (λ, v);
3-3. through fixing v,
Figure FDA0003555673280000017
b, rewriting the formula (1) to a formula (3):
Figure FDA0003555673280000018
the optimal solution of θ obtained by the lagrange multiplier method is shown in formula (4):
Figure FDA0003555673280000019
in the formula (4), θiDescribing the importance degree of each characteristic for describing the ith element of the factor theta;
Figure FDA0003555673280000021
is a variable of
Figure FDA0003555673280000022
The ith element of (1);
thereby further rewriting the objective function formula (1) into formula (5);
Figure FDA0003555673280000023
and 4, obtaining the updating rules of all variables by fixing other variables and reserving only one variable according to the target function shown in the formula (5), and further obtaining the variable in the formula (5)
Figure FDA0003555673280000024
Describing a factor v, a deviation b;
step 5, carrying out electroencephalogram data acquisition on the testee; inputting the obtained electroencephalogram data test set serving as a sample matrix into a computer for determining v, theta,
Figure FDA0003555673280000025
b, acquiring a predicted value label corresponding to the electroencephalogram data in the target function formula (5); and the predicted value label is the detected fatigue degree of the testee.
2. The electroencephalogram fatigue detection method based on the sample and characteristic quality joint quantitative evaluation as claimed in claim 1, characterized in that: in step 2, the subject is gradually fatigued by continuously simulating driving on a straight road by using the simulated driving environment.
3. The electroencephalogram fatigue detection method based on the sample and characteristic quality joint quantitative evaluation as claimed in claim 1, characterized in that: the label vector y in the step 2 is determined by the closing time of eyes of the subject in the electroencephalogram data acquisition process; the label y is represented by a value between 0 and 1.
4. The electroencephalogram fatigue detection method based on the sample and characteristic quality joint quantitative evaluation as claimed in claim 1, characterized in that: v is obtained in step 4,
Figure FDA0003555673280000026
The specific process of b is as follows;
4-1, fixing the steel wire by a fixing device b,
Figure FDA0003555673280000027
To update v such that equation (5) is written as shown in equation (6);
Figure FDA0003555673280000028
in the formula (6)
Figure FDA0003555673280000029
Deriving v in the formula (6) to obtain an update rule of v;
Figure FDA00035556732800000210
wherein liA loss function for a machine learning model expressed as
Figure FDA00035556732800000211
4-2. through fixing
Figure FDA00035556732800000212
v updates b so that equation (5) is written as shown in equation (8);
Figure FDA00035556732800000213
let diagonal matrix
Figure FDA0003555673280000031
So that equation (8) is written as shown in equation (9):
Figure FDA0003555673280000032
taking b in equation (9) as a derivative and letting the derivative be 0, the update rule for b is given by equation (10):
b＝(UTU)-1(UTU)(y-XTw) (10)
4-3. update by fixing v, b
Figure FDA0003555673280000033
Such that equation (10) is written as shown in equation (11);
Figure FDA0003555673280000034
let diagonal matrix
Figure FDA0003555673280000035
So that equation (11) is written as shown in equation (12):
Figure FDA0003555673280000036
to the formula (12)
Figure FDA0003555673280000037
Derivative and let the derivative be 0, get
Figure FDA0003555673280000038
The update rule of (2) is equation (13):
Figure FDA0003555673280000039
5. the electroencephalogram fatigue detection method based on the sample and characteristic quality joint quantitative evaluation as claimed in claim 1, characterized in that: the pretreatment process in step 2 is as follows:
2-1, down-sampling the electroencephalogram data to 200Hz, and performing band-pass filtering on the electroencephalogram data to a range of 1-50 Hz; according to the 5-frequency band method, the method is divided into five frequency bands of Delta, Theta, Alpha, Beta and Gamma
2-2, respectively carrying out short-time Fourier transform with 4 seconds of time window and no overlap on the electroencephalogram data of the 5 frequency bands, and extracting differential entropy characteristics h (X) as shown in a formula (14):
h(X)＝-∫xf(x)ln(f(x))dx (14)
in the formula (14), X is an input sample matrix, and X is an element in the input sample matrix; (x) is a probability density function;
the updated differential entropy characteristic h (X) is shown as a formula (15);
Figure FDA00035556732800000310
in the formula (15), σ is a standard deviation of the probability density function; μ is the expectation of the probability density function.
6. The electroencephalogram fatigue detection method based on the sample and characteristic quality joint quantitative evaluation as claimed in claim 1, characterized in that: 17 leads are adopted for the electroencephalogram data acquisition, and 5 frequency bands are selected; the 5 frequency bands are respectively 1-4Hz, 4-8Hz, 8-14Hz, 14-31Hz and 31-50 Hz.
CN202110317792.5A 2021-03-24 2021-03-24 Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner Active CN113143275B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110317792.5A CN113143275B (en) 2021-03-24 2021-03-24 Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110317792.5A CN113143275B (en) 2021-03-24 2021-03-24 Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner

Publications (2)

Publication Number Publication Date
CN113143275A CN113143275A (en) 2021-07-23
CN113143275B true CN113143275B (en) 2022-04-26

Family

ID=76884692

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110317792.5A Active CN113143275B (en) 2021-03-24 2021-03-24 Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner

Country Status (1)

Country Link
CN (1) CN113143275B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116746931B (en) * 2023-06-15 2024-03-19 中南大学 Incremental driver bad state detection method based on brain electricity

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5846208A (en) * 1996-09-04 1998-12-08 Siemens Aktiengesellschaft Method and apparatus for the evaluation of EEG data
CN101155548A (en) * 2005-03-04 2008-04-02 曼提斯库拉Ehf.公司 A method and a system for assessing neurological conditions
CN105957047A (en) * 2016-05-06 2016-09-21 中国科学院自动化研究所 Supervised multimodal brain image fusion method
CN111067513A (en) * 2019-12-11 2020-04-28 杭州电子科技大学 Sleep quality detection key brain area judgment method based on characteristic weight self-learning

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5846208A (en) * 1996-09-04 1998-12-08 Siemens Aktiengesellschaft Method and apparatus for the evaluation of EEG data
CN101155548A (en) * 2005-03-04 2008-04-02 曼提斯库拉Ehf.公司 A method and a system for assessing neurological conditions
CN105957047A (en) * 2016-05-06 2016-09-21 中国科学院自动化研究所 Supervised multimodal brain image fusion method
CN111067513A (en) * 2019-12-11 2020-04-28 杭州电子科技大学 Sleep quality detection key brain area judgment method based on characteristic weight self-learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《Identifying Stable Patterns over Time for Emotion Recognition from EEG》;Wei-Long Zheng;《 IEEE Transactions on Affective Computing》;20170605;全文 *
《Investigating Critical Frequency Bands and Channels for EEG-Based Emotion Recognition with Deep Neural Networks》;Wei-Long Zheng;《 IEEE Transactions on Autonomous Mental Development 》;20150508;全文 *

Also Published As

Publication number Publication date
CN113143275A (en) 2021-07-23

Similar Documents

Publication Publication Date Title
CN113627518B (en) Method for realizing neural network brain electricity emotion recognition model by utilizing transfer learning
US20200367800A1 (en) Method for identifying driving fatigue based on cnn-lstm deep learning model
CN110070105B (en) Electroencephalogram emotion recognition method and system based on meta-learning example rapid screening
CN108960182B (en) P300 event related potential classification identification method based on deep learning
CN108304917A (en) A kind of P300 signal detecting methods based on LSTM networks
CN106108894A (en) A kind of emotion electroencephalogramrecognition recognition method improving Emotion identification model time robustness
CN112426160A (en) Electrocardiosignal type identification method and device
CN112674782B (en) Device and method for detecting epileptic-like electrical activity of epileptic during inter-seizure period
CN112773378B (en) Electroencephalogram emotion recognition method for feature weight adaptive learning
CN114970599A (en) Identification method and identification device for attention defect associated electroencephalogram signals and storage medium
CN111067513B (en) Sleep quality detection key brain area judgment method based on characteristic weight self-learning
CN111513730A (en) Psychological stress prediction method and system based on multi-channel physiological data
CN115563484A (en) Street greening quality detection method based on physiological awakening identification
CN113143275B (en) Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner
CN114925734A (en) Online neuron classification method based on neural mimicry calculation
CN113576481A (en) Mental load assessment method, device, equipment and medium
CN111568412A (en) Method and device for reconstructing visual image by utilizing electroencephalogram signal
CN113988123A (en) Electroencephalogram fatigue prediction method based on self-weighted increment RVFL network
CN112244767A (en) Cognitive function assessment method based on electroencephalogram
CN113128585B (en) Deep neural network based multi-size convolution kernel method for realizing electrocardiographic abnormality detection and classification
CN115736920A (en) Depression state identification method and system based on bimodal fusion
CN114903495A (en) Headset implementation method capable of monitoring human body pressure
CN112438741A (en) Driving state detection method and system based on electroencephalogram feature transfer learning
Alfaro et al. Continuous neural networks for electroencephalography waveform classification
CN113705649B (en) EMD-SVD feature extraction-based hand tremor detection method and system

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