CN108937921A - In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density - Google Patents
In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density Download PDFInfo
- Publication number
- CN108937921A CN108937921A CN201810320573.0A CN201810320573A CN108937921A CN 108937921 A CN108937921 A CN 108937921A CN 201810320573 A CN201810320573 A CN 201810320573A CN 108937921 A CN108937921 A CN 108937921A
- Authority
- CN
- China
- Prior art keywords
- signal
- mode decomposition
- empirical mode
- spectral density
- eeg signals
- 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.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
-
- 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)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Psychiatry (AREA)
- Public Health (AREA)
- Artificial Intelligence (AREA)
- Psychology (AREA)
- Child & Adolescent Psychology (AREA)
- Educational Technology (AREA)
- Hospice & Palliative Care (AREA)
- Social Psychology (AREA)
- Developmental Disabilities (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses the driving fatigue feature extracting methods of a kind of combination empirical mode decomposition and energy spectral density.The present invention is comprised the steps of: 1, is acquired driving EEG signals using brain wave acquisition equipment;2, collected EEG signals are pre-processed, including frequency reducing, noise reduction;3, to pretreated signal by combining the feature extracting method of empirical mode decomposition and energy spectral density to extract feature;4, classification learning, identification are carried out using classifier to the feature of extraction.The present invention can effectively improve subsequent classification and Detection accuracy rate using combining the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density to extract driving fatigue feature.
Description
Technical field
The present invention relates to the feature extracting method of driving fatigue, in particular to a kind of combination empirical mode decomposition and energy spectrum
The driving fatigue feature extracting method of density.
Background technique
Empirical mode decomposition (Empirical Mode Decomposition, EMD) is Huang E (N.E.Huang) in the U.S.
National Aerospace Bureau and a kind of NEW ADAPTIVE signal time frequency processing method that other people creatively proposed in 1998, it is especially suitable
Analysis for nonlinear and non local boundary value problem is handled, and is the pith of Hilbert-Huang transform.Its time ruler signal-based
Degree generates adaptive basic function, is a series of intrinsic mode function (Intrinsic from high to low by given signal decomposition
Mode Function, IMF) component.Each component has an independent time scale, and each component has orthogonality, complete
The features such as standby property.EMD can be regarded to one kind of wavelet decomposition as, its subband is established according to each component of decomposed signal, is produced
Raw each IMF component represents the details of signal on certain scale and frequency band.
Energy spectrum (Energy Spectral Density, ESD) is one kind of spectrum signature.Dynamic Signal contains because of it
The signal component of multiple frequency ranges is difficult to effectively distinguish the ingredient of signal in the time domain, therefore can be right on frequency domain
Signal is analyzed and processed, and fundamental characteristics one of of the spectral characteristic as signal, is one of main feature of feature extraction.It is logical
Crossing Fourier analysis method can be many harmonic components by dynamic signal decomposition, and wherein each harmonic component is by it
Amplitude and phase characterization, when each harmonic wave by frequency height be arranged in spectrum shape, be formed frequency spectrum.
Summary of the invention
The purpose of the present invention is power spectrum to signal carry out feature extraction on the basis of, in conjunction with particle swarm optimization algorithm
(PSO) norm of learning machine is transfinited to multilayer study and scale factor is iterated optimizing, propose a kind of excellent based on population
The multilayer study of change is transfinited the driving fatigue detection method of learning machine (PSO-HELM).
According to technical solution provided by the invention, the driving of a kind of combination empirical mode decomposition and energy spectral density is proposed
Fatigue characteristic extracting method, includes the following steps:
Step 1 acquires driver's EEG signals using brain wave acquisition equipment;The eeg signal acquisition includes that record is driven
The real-time change of the person's of sailing EEG signals, using the length of 10 seconds as every section EEG signals segment.
Step 2 pre-processes collected EEG signals, including frequency reducing, noise reduction;
Step 3 passes through the feature extraction in conjunction with empirical mode decomposition and energy spectral density to pretreated EEG signals
Method extracts feature;
Step 4 carries out classification learning, identification using classifier to the feature of extraction.
To pretreated EEG signals by combining the feature extracting method of empirical mode decomposition and energy spectral density to mention
Feature is taken, specifically:
Step 3-1: a series of natural mode of vibration components are obtained by empirical mode decomposition to signal after pretreatment;
Step 3-2: to the feature of three first layers intrinsic mode function component extraction energy spectral density, and in this, as signal spy
Sign.
Empirical mode decomposition in the step 3-1 specifically:
(1) all maximum points in EEG signals are identified and are fitted to the coenvelope line e of signalup(t), EEG signals are identified
In all minimum points and be fitted to the lower envelope line e of signallow(t);
(2) according to the line computation average value of the envelope up and down m of synthesis1(t), formula are as follows:
(3) pretreated EEG signals x (t) is subtracted into m1(t) h is obtained1(t), the h that will be obtained1(t) as new brain
Electric signal,
(4) step (1)-(3) are repeated k times, until h1(t) meet;I.e. extreme point number is equal with zero number or phase
Poor one, and the mean value of upper and lower envelope is zero in any point;H at this time1(t) it is first IMF component of signal, is denoted as
c1(t)=h1(t);
(5) by c1(t) with the difference r of x (t)1(t) as new signal;
Step (1)-(5) are repeated, a series of IMF component is obtained, when obtained IMF component or residue signal is less than in advance
When the value of setting, empirical mode decomposition step is completed.
To three first layers intrinsic mode function component extraction energy spectrum signature in the step 3-2 specifically: calculate each
The energy spectral integral of layer intrinsic mode function component, the i.e. energy that each layer of intrinsic mode function component includes, the energy are
The feature of every layer of intrinsic mode function component;The formula of energy spectrum is as follows:
Wherein, fnFor signal component sequence, F (ω) is signal fnFourier transform, F*(ω) is F (ω)
Conjugate function, ωkWhat is indicated is the corresponding angular frequency of signal spectrum, and that n is represented is the sampling number of signal, j
What is indicated is imaginary number.
The present invention has the beneficial effect that:
The spectrum signature of signal is obtained using empirical mode decomposition, and feature extraction and traditional is carried out by energy spectrum
Band power spectrum signature extracting method is compared, and the discrimination of classification and Detection is effectively raised.
Detailed description of the invention
Fig. 1 is empirical mode decomposition schematic diagram;
The selection and extract flow chart that Fig. 2 is characterized.
Specific embodiment:
The present invention is further explained in the light of specific embodiments.It is described below only as demonstration and explanation, not
It is intended that the present invention is limited in any way.
The step of present invention realizes as shown in Figures 1 and 2 is as follows:
Step 1 acquires driving EEG signals using brain wave acquisition equipment.Wherein collected EEG signals be originated from 6 by
The international 10-20 lead system electrode cap that examination person wears, port number are 32 channels, frequency 1000Hz.It records 20 minutes respectively
EEG signals when normal driving and fatigue driving in 20 minutes, using every 10 seconds signals as sample, the total acquisition of 6 subjects
1440 samples.Wherein using 480 samples as training sample, 960 brain electricity samples are test sample;
Step 2 pre-processes collected EEG signals, including frequency reducing, noise reduction.Pretreated signal sampling frequency
Rate is reduced to 200Hz, primary frequency range 0.1Hz-50Hz.The sampled point of each sample becomes 2000 from 10000;
Step 3 passes through the feature extracting method in conjunction with empirical mode decomposition and energy spectral density to pretreated signal
Extract feature;
Step 4 carries out classification learning, identification using classifier to the feature of extraction, uses support vector machines in this example
(SVM), the Multilayer Perception of k nearest neighbor (KNN) and the particle group optimizing learning machine (PSO-H-ELM) that transfinites carries out comparison of classification respectively;
In the step 1, eeg signal acquisition include record driver's EEG signals real-time change, using 10 seconds as
The length of every section of EEG signals segment, and every section of EEG signals segment is analyzed and processed.
In the step 3, the feature in conjunction with empirical mode decomposition and energy spectral density is passed through to pretreated signal
Extracting method extracts the step of feature specifically:
Step 3-1: a series of natural mode of vibration components are obtained by empirical mode decomposition to signal after pretreatment;
Step 3-2: to three first layers intrinsic mode function component extraction energy spectrum signature, and in this, as signal characteristic.
In the step 3, a series of natural mode of vibration components are obtained by empirical mode decomposition to signal after pretreatment
Step specifically:
(1) all maximum points in EEG signals are identified and are fitted to the coenvelope line e of signalup(t), EEG signals are identified
In all maximum points and be fitted to the lower envelope line e of signallow(t);
(2) according to the line computation average value of the envelope up and down m of synthesis1(t), formula are as follows:
(3) x (t) is subtracted into m1(t) h is obtained1(t), the h that will be obtained1(t) as new EEG signals x (t), step is repeated
(1), it is screened by k times, until h1(t) meet: extreme point number is equal with zero number or most differences one, and upper and lower
The mean value of envelope is zero in any point.H at this time1(t) it is first IMF component of signal, is denoted as c1(t)=h1(t);
(4) by c1(t) with the difference r of x (t)1(t)=x (t)-c1(t) it repeats the above steps as new signal, obtains one
The IMF component of series, when obtained IMF component or residue signal is less than preset value, empirical mode decomposition step is complete
At.
In the step 3, to three first layers intrinsic mode function component extraction energy spectrum signature, and in this, as signal spy
The step of sign is classified, is identified specifically: calculate the energy spectral integral of each layer of intrinsic mode function component, i.e., each layer is solid
The energy for having mode function component to include, the energy are the feature of every layer of intrinsic mode function component.The formula of energy spectrum is such as
Under:
Wherein, fnFor signal component sequence, F (ω) is signal fnFourier transform, F*(ω) is the conjugation letter of F (ω)
Number, ωkWhat is indicated is the corresponding angular frequency of signal spectrum, and what n was represented is the sampling number of signal, and what j was indicated is imaginary number.
Feature is carried out using the method and traditional frequency band energy spectral method that combine empirical mode decomposition and energy spectrum respectively
It extracts, classification results are as shown in table 1 below.
1 four kinds of sorting algorithm classification accuracy comparisons of table
By the Classification and Identification rate for comparing three kinds of algorithms, it can be clearly seen that using identical sorting algorithm
There is better classifying quality than traditional frequency band energy spectral method in conjunction with the method for empirical mode decomposition and energy spectrum, is using
In the case where PSO-H-ELM algorithm, average classification accuracy has reached 94.10%.Show to combine empirical mode decomposition and energy
The method of spectrum has great advantage in the detection and analysis application of driving fatigue EEG signals.
Claims (5)
1. combining the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density, it is characterised in that including walking as follows
It is rapid:
Step 1 acquires driver's EEG signals using brain wave acquisition equipment;
Step 2 pre-processes collected EEG signals, including frequency reducing, noise reduction;
Step 3 passes through the feature extracting method in conjunction with empirical mode decomposition and energy spectral density to pretreated EEG signals
Extract feature;
Step 4 carries out classification learning, identification using classifier to the feature of extraction.
2. the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density is combined as described in claim 1,
It is characterized in that, step 1 specifically: the eeg signal acquisition includes the real-time change for recording driver's EEG signals, with 10
Length of the second as every section of EEG signals segment.
3. the driving fatigue feature extracting method of combination empirical mode decomposition according to claim 1 and energy spectral density,
It is characterized in that, step 3 specifically:
Step 3-1: a series of natural mode of vibration components are obtained by empirical mode decomposition to signal after pretreatment;
Step 3-2: to the feature of three first layers intrinsic mode function component extraction energy spectral density, and in this, as signal characteristic.
4. the driving fatigue feature extracting method of combination empirical mode decomposition according to claim 1 and energy spectral density,
It is characterized in that empirical mode decomposition in the step 3-1 specifically:
(1) all maximum points in EEG signals are identified and are fitted to the coenvelope line e of signalup(t), institute in EEG signals is identified
There is minimum point and is fitted to the lower envelope line e of signallow(t);
(2) according to the line computation average value of the envelope up and down m of synthesis1(t), formula are as follows:
(3) pretreated EEG signals x (t) is subtracted into m1(t) h is obtained1(t), the h that will be obtained1(t) as new brain telecommunications
Number,
(4) step (1)-(3) are repeated k times, until h1(t) meet;I.e. extreme point number it is equal with zero number or difference one
It is a, and the mean value of upper and lower envelope is zero in any point;H at this time1(t) it is first IMF component of signal, is denoted as c1(t)
=h1(t);
(5) by c1(t) with the difference r of x (t)1(t) as new signal;
Step (1)-(5) are repeated, a series of IMF component is obtained, are preset when obtained IMF component or residue signal is less than
Value when, empirical mode decomposition step complete.
5. the driving fatigue feature extracting method of combination empirical mode decomposition according to claim 1 and energy spectral density,
It is characterized in that, to three first layers intrinsic mode function component extraction energy spectrum signature in the step 3-2 specifically: calculate every
The energy that the energy spectral integral of one layer of intrinsic mode function component, i.e. each layer of intrinsic mode function component include, the energy are
For the feature of every layer of intrinsic mode function component;The formula of energy spectrum is as follows:
Wherein, fnFor signal component sequence, F (ω) is signal fnFourier transform, F*(ω) is the conjugate function of F (ω), ωk
What is indicated is the corresponding angular frequency of signal spectrum, and what n was represented is the sampling number of signal, and what j was indicated is imaginary number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810320573.0A CN108937921A (en) | 2018-04-11 | 2018-04-11 | In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810320573.0A CN108937921A (en) | 2018-04-11 | 2018-04-11 | In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108937921A true CN108937921A (en) | 2018-12-07 |
Family
ID=64498765
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810320573.0A Pending CN108937921A (en) | 2018-04-11 | 2018-04-11 | In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108937921A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112508088A (en) * | 2020-12-03 | 2021-03-16 | 重庆邮智机器人研究院有限公司 | DEDBN-ELM-based electroencephalogram emotion recognition method |
CN113180705A (en) * | 2021-04-07 | 2021-07-30 | 北京脑陆科技有限公司 | Fatigue detection method and system based on EEG brain waves |
CN113221843A (en) * | 2021-06-07 | 2021-08-06 | 吉林大学 | Driving style classification method based on empirical mode decomposition characteristics |
CN113288150A (en) * | 2021-06-25 | 2021-08-24 | 杭州电子科技大学 | Channel selection method based on fatigue electroencephalogram combination characteristics |
CN113633288A (en) * | 2021-06-25 | 2021-11-12 | 杭州电子科技大学 | EEMD-based fatigue electroencephalogram feature extraction method |
CN114813934A (en) * | 2022-03-02 | 2022-07-29 | 西安交通大学 | Method for detecting and identifying surface material of target object in robot environment |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140316230A1 (en) * | 2013-04-22 | 2014-10-23 | Personal Neuro Devices Inc. | Methods and devices for brain activity monitoring supporting mental state development and training |
CN105105774A (en) * | 2015-10-09 | 2015-12-02 | 吉林大学 | Driver alertness monitoring method and system based on electroencephalogram information |
CN105677035A (en) * | 2016-01-26 | 2016-06-15 | 杭州电子科技大学 | EEMD (Ensemble Empirical Mode Decomposition) and wavelet threshold based motor imagery electroencephalogram signal denoising method |
CN107361766A (en) * | 2017-07-17 | 2017-11-21 | 中国人民解放军信息工程大学 | A kind of mood EEG signal identification method based on EMD domains multidimensional information |
-
2018
- 2018-04-11 CN CN201810320573.0A patent/CN108937921A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140316230A1 (en) * | 2013-04-22 | 2014-10-23 | Personal Neuro Devices Inc. | Methods and devices for brain activity monitoring supporting mental state development and training |
CN105105774A (en) * | 2015-10-09 | 2015-12-02 | 吉林大学 | Driver alertness monitoring method and system based on electroencephalogram information |
CN105677035A (en) * | 2016-01-26 | 2016-06-15 | 杭州电子科技大学 | EEMD (Ensemble Empirical Mode Decomposition) and wavelet threshold based motor imagery electroencephalogram signal denoising method |
CN107361766A (en) * | 2017-07-17 | 2017-11-21 | 中国人民解放军信息工程大学 | A kind of mood EEG signal identification method based on EMD domains multidimensional information |
Non-Patent Citations (2)
Title |
---|
周炯槃 等: "《通信原理(第4版)》", 31 August 2015, 北京邮电大学出版社 * |
张学军 等: "公共空间模式算法结合经验模式分解的EEG特征提取", 《计算机工程与应用》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112508088A (en) * | 2020-12-03 | 2021-03-16 | 重庆邮智机器人研究院有限公司 | DEDBN-ELM-based electroencephalogram emotion recognition method |
CN113180705A (en) * | 2021-04-07 | 2021-07-30 | 北京脑陆科技有限公司 | Fatigue detection method and system based on EEG brain waves |
CN113221843A (en) * | 2021-06-07 | 2021-08-06 | 吉林大学 | Driving style classification method based on empirical mode decomposition characteristics |
CN113288150A (en) * | 2021-06-25 | 2021-08-24 | 杭州电子科技大学 | Channel selection method based on fatigue electroencephalogram combination characteristics |
CN113633288A (en) * | 2021-06-25 | 2021-11-12 | 杭州电子科技大学 | EEMD-based fatigue electroencephalogram feature extraction method |
CN114813934A (en) * | 2022-03-02 | 2022-07-29 | 西安交通大学 | Method for detecting and identifying surface material of target object in robot environment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108937921A (en) | In conjunction with the driving fatigue feature extracting method of empirical mode decomposition and energy spectral density | |
CN108375472A (en) | Based on the Method for Bearing Fault Diagnosis and system and device for improving experience wavelet transformation | |
CN105510687B (en) | A kind of electric voltage exception characteristic recognition method based on empirical mode decomposition | |
CN106803081A (en) | A kind of brain electricity sorting technique based on Multi-classifers integrated | |
CN110163128A (en) | The Method of EEG signals classification of improved EMD algorithm combination wavelet package transforms and CSP algorithm | |
CN108042132A (en) | Brain electrical feature extracting method based on DWT and EMD fusions CSP | |
CN111476339B (en) | Rolling bearing fault feature extraction method, intelligent diagnosis method and system | |
CN109238182B (en) | Fourier spectrum characteristic-based objective fabric flatness rating method | |
Kumar et al. | Classification of seizure and seizure-free EEG signals using multi-level local patterns | |
CN111553207A (en) | Statistical distribution-based ship radiation noise characteristic recombination method | |
CN102982347A (en) | Method for electric energy quality disturbance classification based on KL distance | |
CN113116361A (en) | Sleep staging method based on single-lead electroencephalogram | |
CN116153329A (en) | CWT-LBP-based sound signal time-frequency texture feature extraction method | |
CN114139586A (en) | Electric energy quality composite disturbance identification method based on multi-scale analysis and two-dimensional feature extraction | |
Mahla et al. | Recognition of complex and multiple power quality disturbances using wavelet packet-based fast kurtogram and ruled decision tree algorithm | |
CN109254654A (en) | A kind of driving fatigue feature extracting method of combination PCA and PCANet | |
CN108108659B (en) | Island detection key feature extraction method based on empirical mode decomposition | |
CN114036977A (en) | Fan power quality disturbance monitoring method based on two-dimensional image coding features | |
Shinde et al. | Gender classification with KNN by extraction of Haar wavelet features from canny shape fingerprints | |
CN117421634A (en) | Electromagnetic interference source rapid identification method and device based on spectrum joint image analysis | |
CN116911189A (en) | Method for precisely detecting dynamic time-varying direct current fault arc by applying continuous learning | |
Khoje et al. | Comparative performance evaluation of fast discrete curvelet transform and colour texture moments as texture features for fruit skin damage detection | |
CN106443182B (en) | A kind of method that test item Electromagnetic Launching amplitude spectrum data based on spectrum signature are classified | |
CN115581467A (en) | SSVEP (steady state visual evoked potential) identification method based on time, frequency and time-frequency domain analysis and deep learning | |
CN115498956A (en) | Photovoltaic array series arc fault diagnosis method |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181207 |