CN107440716A - Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal - Google Patents
Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal Download PDFInfo
- Publication number
- CN107440716A CN107440716A CN201710617410.4A CN201710617410A CN107440716A CN 107440716 A CN107440716 A CN 107440716A CN 201710617410 A CN201710617410 A CN 201710617410A CN 107440716 A CN107440716 A CN 107440716A
- Authority
- CN
- China
- Prior art keywords
- mtd
- msub
- mrow
- walking
- mtr
- 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
- A61B5/389—Electromyography [EMG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/112—Gait analysis
-
- 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
-
- 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
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Abstract
The invention belongs to the technical field of the processing of non-stationary nonlinear surface electromyographic signal and the lower extremity movement classification of motion, and in particular to a kind of human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal.Electromyographic signal of this method acquired in using the single channel obtained at a certain single lower limb muscle group related to the action of daily lower extremity movement, obtain different subjects, leg extension, leg curvature, walking support and walking swing electromyographic signal data corresponding to four kinds of typical lower extremity movement operating states, then wavelet transformation is carried out to above-mentioned signal and obtains wavelet coefficient, singular value decomposition is carried out to each layer of wavelet coefficient as eigenmatrix.Four classification problems are finally changed into by multiple two classification problems using the tree-shaped grader of SVMs two, obtain identification result of classifying.The result is shown under clinical testing data, feature extraction and classifying method of the invention, to the single channel electromyographic signal of four kinds of daily lower extremity movement actions, has good effect.
Description
Technical field
The invention belongs to the technical field of the processing of non-stationary nonlinear surface electromyographic signal and the lower extremity movement classification of motion, tool
Body is related to a kind of human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal.
Background technology
Human body limb movement is the important component in mankind's ADL, and for special population, (such as hemiplegia is suffered from
Person or older population etc.) ADL (especially lower extremity movement is such as walked, stands, squats down or sat back and waited), generally require
External auxiliary can just smoothly complete.A large amount of scholars are working on the research about power-assisted robot system, above-mentioned to meet
The demand of the daily life campaign auxiliary of crowd.Because surface electromyogram signal (EMG) can directly represent ADL
Electrophysiologic response, therefore most of this kind of system is as signal source using EMG signal.
Electromyographic signal is a kind of nonlinear electro-physiological signals of complicated non-stationary, can characterize human muscle's contraction process
The physiological activity of middle neuromuscular system.The relation of biofeedback and lower extremity movement based on EMG signal is power-assisted machine
The emphasis of device people system concern.It is therefore proposed that a kind of sorting technique based on the action of EMG signal lower extremity movement of robust is very
It is necessary.
Moving cell generally includes the skeletal muscle fibre of motor neuron and motor neurons innervate.The strength of contraction of muscle
It is to be adjusted by the quantity of the moving cell activated, and the EMG signal measured at any single muscle group obtained is by more
The activation of individual moving cell and substantial amounts of noise are superimposed what is formed.Therefore, it is difficult there is serious noise pollution in EMG signal
Directly to be classified with it.Traditional method is that different knee fortune is detected using the time and frequecy characteristic of EMG signal
Dynamic model formula, such as average absolute value (MAV), root mean square (RMS), zero-crossing rate (ZC), average frequency (MNF) and median frequency
(MDF).The frequency of EMG signal changes over time, and traditional analysis method can not describe its time dependence exactly,
It is follow-up to have researched and proposed some EMG signal Feature Selection Models in order to solve this problem, including Fourier transformation (FT),
Wavelet transformation (WT), autoregression (AR) and power Spectral Estimation (PSE).Wavelet transformation considers the time and frequency domain characteristics of signal, its
Classification accuracy in most cases is all more contour than Fourier transformation.Many studies have shown that the small echo in EMG feature extractions
Analysis has had and has been widely applied in many advantages in terms of pattern-recognition and in the classification of EMG motion states, but seldom
Have and be used in the motion of knee joint state classification of single EMG passages.
The Chinese patent of Application No. 201610326948.5 discloses a kind of for human body lower limbs surface electromyogram signal
Discrimination method, the method discloses the preprocess method of electromyographic signal (including Hz noise filtering, baseline drift filtering and Gauss
Three kinds of methods of white noise filter), and feature extraction algorithm (wavelet transformation and singular value decomposition are combined).The patent application carries
The discrimination method of the electromyographic signal gone out, does not combine closely with practical application, and it is single channel that can not be directly used in signal source
The human body lower limbs athletic performance classification identification of lower electromyographic signal.
The content of the invention
A kind of purpose of the present invention, aiming above mentioned problem, there is provided feature based on wavelet transformation and singular value decomposition
Extracting method, for (walking support, walking to be swung, leg extension to different lower extremity movement actions under the conditions of single channel EMG signal
And leg curvature) classified.The EMG signal that this method uses derives from the muscle of preaxial flesh, for from all lower limb
The EMG signal obtained at muscle group tissue.Wavelet transformation and singular value decomposition are feature extracting methods important in pattern-recognition,
But the method that wavelet transformation and singular value decomposition combine is used for four kinds of lower extremity movement actions under the conditions of single channel EMG signal
(walking support, walking swing, stand and squat down) classification be present invention firstly provides.
The technical scheme is that：Human motion classification of motion discrimination method based on single channel electromyographic signal, it is special
Sign is, comprises the following steps：
S1, data acquisition：
Human body electromyographic signal at lower limb muscles group under special exercise operation condition is caught by single myoelectricity passage, obtained
One single pass myoelectricity data；The myoelectricity data are the vectorial S (t) on a time dimension, S (t)=(x1,
x2...xn)；
It should be strongly noted that the Data Collection proposed in this programme is tested by UTS
(University of Technology Sydney, UTS) human research Ethics Committee ratifies, including 14 health and not
Trained subject carries out three experimentations related to motion of knee joint, leg extension (referred to as " seat-stand "), leg
Portion's bending (referred to as "-seat of standing "), walking support and walking are swung, while are monitored and recorded from the inboard leg muscle detected
EMG signal, the sample rate of signal is 1kHz.Experiment is carried out under two independent experimental paradigms, i.e., " stand up-sit down " tests
Normal form and " walking " experimental paradigm.It is tested random these motion processes of progress.The each action of requirement of experiment subject is kept for one second
Movement space, have time of having a rest of five seconds between each two action.The carry out five times of motion repeatedly every time, avoid subject because with
Experiment caused by chance error difference is lack of standardization.For " stand up-sit down ", it is tested random slave station or sits setting in motion.In addition, this is not
With the time of having a rest for having 15 minutes between motion stage.In experimentation, it is proposed that a kind of method of tone timing, the timing
Automatic tone is followed, the beginning that is acted in each motion starts to match with listening Buddhaghosa, to remind it to start and to stop its fortune
It is dynamic.By two motion stages, we obtain the EMG data sets of four types.
S2, the myoelectricity data to acquisition pre-process：
S21, the high-pass filter for being 20Hz with cut-off frequency are filtered to the myoelectricity data that step S1 is obtained；
S22, human body lower limbs athletic performance is divided into leg extension, leg curvature, walking support and walking four classes of swing, obtained
Take the single channel myoelectricity data of above-mentioned four classes action；
S23, repeat the above steps the myoelectricity data for obtaining multiple targets under described four different actions, and by all phases
With the lower data segment obtained of action as one group of data, the data of four class operating states are obtained, the data can characterize above-mentioned four class
Human body lower limbs athletic performance, there is significant difference；
S3, wavelet transformation：
The sample of every group of data to being obtained in step S22 applies Pyatyi Wavelet Transformation Algorithm respectively, be decomposed into cD1,
CD2, cD3, cD4, cD5 and cA5, the corresponding frequency band of each component is 256~512Hz, 128~256Hz, 64~128Hz, 32~
64Hz, 16~32Hz；CAn, n=1,2 ..., 5, representative be signal low-frequency component, cDm, m=1,2 ..., n, representative
It is the radio-frequency component of signal；
S4, singular value decomposition：
Singular value decomposition is carried out to each layer of wavelet coefficient obtained in step S3, compresses it into a characteristic parameter,
To obtain more simple and effective feature input, the eigenmatrix finally obtained is：
Wherein, i is the sample number of athletic performance, and j is the intrinsic dimensionality that each sample obtains；
S5, using the eigenmatrix obtained in step S4 as sample, feature samples are trained using SVMs, and
Support vector machine classifier is generated to be used to recognize the human motion classification of motion.
Most of EMG signal sorting techniques all only considered time domain or the information of frequency domain, the letter of the feature so obtained
Cease relatively fewer, it is impossible to there is good classifying quality, and the wavelet transformation based on time-frequency domain information can be by signal decomposition into not
With yardstick, and from time domain and the more information of frequency domain extraction.Singular value decomposition is extracting important characteristic aspect with fine
Effect.A kind of method that the present invention proposes wavelet transformation and singular value decomposition is combined.Extract each motion state signal
Time and frequency domain information, singular value decomposition then is carried out to each layer of wavelet coefficient, obtains can be used for the spy effectively to classify
Sign.More classification problems can regard multiple two classification problems as, and we used the SVMs (Support of classics
Vector Machine, SVM) grader classified.
Further, in the step S5, generation support vector machine classifier is used to recognize the human motion classification of motion
Specific method be：Human motion action is first divided into by two classes of standing and walking using the first support vector machine classifier, stood
Class includes leg extension and leg curvature, and walking class includes walking support and walking is swung；Again using the second SVMs point
Class device recognizes to standing class, distinguishes leg extension (from station is sat on) and leg curvature (slave station to seat) two classes action, uses
3rd support vector machine classifier recognizes to walking class, distinguishes walking support and walking swings the action of two classes.
Further, in the step S1, single pass electromyographic signal is specifically defined as：Moved with daily lower extremity movement
Make (including leg extension (from station is sat on), leg curvature (slave station to sit), walking support and walking swing) related a certain list
The surface electromyogram signal obtained at one lower limb limb muscle group；The lower limb muscles group related to the action of daily lower extremity movement includes：
Long adductor muscle, gracilis, rectus femoris, musculus vastus lateralis, vastus medialis, sartorius, biceps muscle of thigh, semitendinosus, semimembranosus, extensor hallucis longus,
Musculus extensor digitorum longus pedis, tibialis anterior, musculus peroneus longus, musculus peroneus brevis.
In the step S21, the preprocess method of electromyographic signal is specially：(1) high pass for being 20Hz with cut-off frequency is filtered
Ripple device is filtered to the myoelectricity data that step S1 is obtained；(2) can also use as in Patent No. 201610326948.5
Electromyographic signal preprocess method disclosed in state's patent, including Hz noise filtering, baseline drift filtering and white Gaussian noise filter
Three kinds of methods of ripple.Three kinds of filtering methods and 20Hz high-pass filtering method can be used in any combination according to signal quality condition, and
Without sequencing requirement.
In the step S3, Wavelet Transformation Algorithm is specifically not limited to five layers of decomposition, also may include two layers or more small echos
Decomposed component.
Beneficial effects of the present invention are the feature extraction side that the present invention is combined based on wavelet transformation and singular value decomposition
Method, according to the wavelet transformation-singular value features for the electromyographic signal extracted, be subsequently used under the conditions of single channel EMG signal not
Classified with lower extremity movement action (walking support, walking swing, leg extension and leg curvature).The EMG that this method uses
Signal derives from the muscle of musculus lateralis interni, because it is mainly related to muscular fatigue, and has to the research of man-machine interface important
Meaning.Wavelet transformation and singular value decomposition are all feature extracting methods important in pattern-recognition, and are classified in EMG motion states
In achieve good effect, but the method that combines of wavelet transformation and singular value decomposition in lower limb exercise classification also without very
Good relevant report.This method first according to the data of angular transducer passage as reference, it is artificial to motion state signal
Divided, then carrying out 5 layers of wavelet transformation to signal obtains wavelet coefficient, and carrying out SVD to each layer of wavelet coefficient decomposes work
For the feature of EMG signal.SVMs (SVM) is a kind of grader for maximizing class interval, can be to linear and non-thread
Property data are classified well, and method of the invention is changed into more classification problems using the tree-shaped graders of SVM bis- multiple
Two classification problems.
Compared with conventional art, the present invention also has the following advantages that：
(1) further directed to specific daily lower extremity movement action, (walking support, walking are swung the present invention, leg extension
(from station is sat on) and leg curvature (slave station to seat)) more classification identifications are carried out, a kind of four classification based on SVMs are disclosed
Method；
(2) electromyographic signal under the present invention further limits signal source as single channel, acquired electromyographic signal are and day
Normal lower extremity movement action (including leg extension (from station is sat on), leg curvature (slave station to seat), walking support and walking are swung)
The surface electromyogram signal obtained at related a certain single lower limb muscle group, the benefit of single channel are to maximize to reduce
The contact of equipment and human body, there is larger application prospect.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention；
Embodiment
Below in conjunction with the accompanying drawings, technical scheme is described in detail：
The present invention has carried out lower limb exercise electromyographic signal data acquisition system first.Share 14 it is healthy and not trained
Subject participate in experiment, be tested before data acquisition and all obtain informed consent form.In data acquisition, by an electricity
Pole is placed on inner side muscle groups and places angular instrument on knee joint.Datalog equipment MWX8 (http://
Www.biometricsltd.com/datalog.htm) it is used for electromyographic signal data acquisition, it includes 8 digital channels and 4
Analog channel.One digital channel is used for the collection for monitoring and recording electromyographic signal, and another is used to record the knee pass detected
The angle measurement data of section.Data are obtained and stored in MWX8 internal storages by microSD cards first.Then fitted using bluetooth
Orchestration is carried out data transmission by real-time Datalog softwares, the off-line analysis for the present invention.Sample rate is 1KHz.For minimum
Impedance between polarizing electrode and skin, before experiment, by the hair on target electrode placement location and dead skin cells from skin table
Shave light in face.After shaving light, with alcohol washes skin, electrode is placed on purported skin after its drying.This experimental electrode is using double
Pole electrode (SX230), it provides an overall electrode with distance between fixed electrode, can obtain continuous high quality knot
Fruit, and effectively limit the risk of crosstalk between electrode.The input impedance of EMG amplifiers is more than 10,000,000M ohms.
Subject carries out three experimentations related to motion of knee joint, leg extension (being referred to as " from station is sat on "), leg
Portion's bending (being referred to as " slave station to seat ") and swing (referred to as " walking " and " support "), while monitor and record from the leg detected
Inner side muscle EMG signal.Experiment has two motion stages, stands up-sits down and walks.Being tested random progress, these are moved
Process.The each action of requirement of experiment subject keeps the movement space of one second, there is the time of having a rest of five seconds between each two action.Often
The carry out of secondary motion repeatedly five times, subject is avoided to be tested caused by some individual factors lack of standardization.For fortune of standing up-sit down
It is dynamic, it is tested random slave station or sits setting in motion.In addition, there is the time of having a rest of 15 minutes between this different motion stage.
In experimentation, it is proposed that a kind of method of tone timing, this regularly follows automatic tone, the beginning acted in each motion with
Buddhaghosa is listened to start to match, to remind it to start and to stop its motion.By two motion stages, we obtain four species
The EMG data sets of type.
In order to verify wavelet transformation proposed by the invention and feature extraction that singular value decomposition is combined and tree-shaped more
Performance of the SVM classifier of classifying in being classified based on single pass lower limb exercise, and improve the robustness of disaggregated model and right
The generalization ability of difference subject data.In pretreatment stage, Butterworth of the original electromyographic signal by cut-off frequency for 20Hz
High-pass filter.We ignore subject information, and according to angle measurement data, the operating state electromyographic signal for representing different is divided
Section, the electromyographic signal section of the expression same action of all subjects is stored in same data set, as same class sample.In
It is that we have respectively obtained the data set for characterizing the action of four classes.It is small using Pyatyi in order to obtain the time and frequency domain characteristics of electromyographic signal
Signal is divided into cD1, cD2, cD3, cD4, cD5 and cA5 by Wave Decomposition, and the corresponding frequency band of each component is 256~512Hz, 128~
256Hz, 64~128Hz, 32~64Hz, 16~32Hz.CAn (n=1,2 ..., 5) represent be signal low frequency component, cDn
(n=1,2 ..., n) what is represented is the high fdrequency component of signal.The singular value of each component, table are calculated by singular value decomposition afterwards
1 shows the average value and variance yields of the singular value of each component.Variance analysis is carried out based on these results, shows lower extremity movement
In this six compositions have it is obvious different.
The average value and variance yields of the singular value of 1 each component of table
Table 3.The mean and STD results of time-frequency features by WT-
based SVD approach basde on experimental tralls from all participants on the
vastus medialis muscle.
In order to which the feature of the motion state different to four classes is classified, we are using tree-shaped based on Radial basis kernel function
Support vector machine classifier.SVM classifier differentiation at root node is stood up-sat down and the two motion stages of walking, two
SVM at leaf node distinguish respectively from sit on station, slave station to sit and walking, support.Count the wrong sample of all SVM classifiers
This, as total error rate.Experiment altogether have collected 260 motion segments, wherein from sit on station 40 times, by station to seat 40 times,
Walking 90 times, support 90 times.In order that acquired results are more stable, the present invention is by the way of 5 folding cross validations, and by iteration
The average result of 50 times is as final result.Specifically, sample is divided into 5 groups by random, retains a subset as checking number
According to remaining four subsets are used to train, and five sons, which are concentrated, will be used to verify data.Based on this method, what is obtained is final
Classification accuracy be 91.85% ± 0.88%..
Claims (3)
1. the human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal, it is characterised in that including following step
Suddenly：
S1, data acquisition：
Human body electromyographic signal at lower limb muscles group under special exercise operation condition is caught by single myoelectricity passage, obtains one
Single pass myoelectricity data；The myoelectricity data are the vectorial S (t) on a time dimension, S (t)=(x1,x2...xn)；
S2, the single channel myoelectricity data to acquisition pre-process：
S21, the high-pass filter for being 20Hz with cut-off frequency are filtered to the myoelectricity data that step S1 is obtained；
S22, human body lower limbs athletic performance is divided into leg extension, leg curvature, walking support and walking swings four classes, in acquisition
State the single channel myoelectricity data of four classes action；
S23, repeat the above steps the myoelectricity data for obtaining multiple targets under described four different actions, and will be all identical dynamic
Make the lower data segment obtained as one group of data, obtain the data of four class operating states, it is anthropoid that the data can characterize above-mentioned four
Lower extremity movement acts, and has significant difference；
S3, wavelet transformation：
The sample of every group of data to being obtained in step S22 applies Pyatyi Wavelet Transformation Algorithm respectively, be decomposed into cD1, cD2,
CD3, cD4, cD5 and cA5, the corresponding frequency band of each component is 256~512Hz, 128~256Hz, 64~128Hz, 32~
64Hz, 16~32Hz；CAn, n=1,2 ..., 5, representative be signal low-frequency component, cDm, m=1,2 ..., n, representative
It is the radio-frequency component of signal；
S4, singular value decomposition：
Singular value decomposition is carried out to each layer of wavelet coefficient obtained in step S3, compresses it into a characteristic parameter, with
Inputted to more simple and effective feature, the eigenmatrix finally obtained is：
<mrow>
<msub>
<mi>M</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>&sigma;</mi>
<mn>11</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mn>12</mn>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mrow>
<mn>1</mn>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>&sigma;</mi>
<mn>21</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mn>22</mn>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mrow>
<mn>2</mn>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>&sigma;</mi>
<mrow>
<mi>i</mi>
<mn>1</mn>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mrow>
<mi>i</mi>
<mn>2</mn>
</mrow>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>&sigma;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, i is the sample number of athletic performance, and j is the intrinsic dimensionality that each sample obtains；
S5, using the eigenmatrix obtained in step S4 as sample, feature samples are trained using SVMs, and generates
Support vector machine classifier is used to recognize the human motion classification of motion.
2. the human body lower limbs athletic performance classification discrimination method according to claim 1 based on single channel electromyographic signal, its
It is characterised by, in the step S5, generation support vector machine classifier is used for the tool recognized of classifying to human body lower limbs athletic performance
Body method is：Human body lower limbs athletic performance is first divided into by two classes of standing and walking using the first support vector machine classifier, stood
Class includes leg extension and leg curvature, and walking class includes walking support and walking is swung；Again using the second SVMs point
Class device recognizes to standing class, leg extension and the action of the class of leg curvature two is distinguished, using the 3rd support vector machine classifier
Walking class is recognized, walking support is distinguished and walking swings the action of two classes.
3. the human body lower limbs athletic performance classification discrimination method according to claim 1 or 2 based on single channel electromyographic signal,
Characterized in that, in the step S1, the definition of single pass electromyographic signal is：Related to the action of daily lower extremity movement is a certain
The surface electromyogram signal obtained at single lower limb muscle group；The lower limb muscles group bag related to the action of daily lower extremity movement
Include：Long adductor muscle, gracilis, rectus femoris, musculus vastus lateralis, vastus medialis, sartorius, biceps muscle of thigh, semitendinosus, semimembranosus, length are stretched
Flesh, musculus extensor digitorum longus pedis, tibialis anterior, musculus peroneus longus, musculus peroneus brevis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710617410.4A CN107440716A (en) | 2017-07-26 | 2017-07-26 | Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710617410.4A CN107440716A (en) | 2017-07-26 | 2017-07-26 | Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107440716A true CN107440716A (en) | 2017-12-08 |
Family
ID=60487508
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710617410.4A Pending CN107440716A (en) | 2017-07-26 | 2017-07-26 | Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107440716A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108388855A (en) * | 2018-02-11 | 2018-08-10 | 宁波工程学院 | A kind of pin electrode EMG Signal Decomposition Based method |
| CN109446972A (en) * | 2018-10-24 | 2019-03-08 | 电子科技大学中山学院 | Gait recognition model establishing method, recognition method and device based on electromyographic signals |
| CN109498370A (en) * | 2018-12-26 | 2019-03-22 | 杭州电子科技大学 | Joint of lower extremity angle prediction technique based on myoelectricity small echo correlation dimension |
| CN109498009A (en) * | 2018-12-26 | 2019-03-22 | 杭州电子科技大学 | Human body daily behavior myoelectricity feature selection approach based on feature class separability index |
| CN109512390A (en) * | 2018-10-26 | 2019-03-26 | 武汉大学 | Sleep stage method and wearable device based on EEG time domain various dimensions feature and M-WSVM |
| CN110414619A (en) * | 2019-08-05 | 2019-11-05 | 重庆工商职业学院 | One kind being based on EMG signal recognition methods |
| CN110672090A (en) * | 2019-09-26 | 2020-01-10 | 深圳大学 | Indoor and outdoor integrated positioning method and system based on electromyographic data |
| CN114344077A (en) * | 2021-12-07 | 2022-04-15 | 华南理工大学 | Flexible upper limb rehabilitation robot system based on SEMG movement intention recognition |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105997064A (en) * | 2016-05-17 | 2016-10-12 | 成都奥特为科技有限公司 | Method for identifying human lower limb surface EMG signals (electromyographic signals) |
-
2017
- 2017-07-26 CN CN201710617410.4A patent/CN107440716A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105997064A (en) * | 2016-05-17 | 2016-10-12 | 成都奥特为科技有限公司 | Method for identifying human lower limb surface EMG signals (electromyographic signals) |
Non-Patent Citations (1)
| Title |
|---|
| 张羿，等: "基于SVM 的下肢运动表面肌电信号的特征提取与辨识分析", 《载人航天》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108388855A (en) * | 2018-02-11 | 2018-08-10 | 宁波工程学院 | A kind of pin electrode EMG Signal Decomposition Based method |
| CN108388855B (en) * | 2018-02-11 | 2022-02-11 | 宁波工程学院 | Needle electrode electromyographic signal decomposition method |
| CN109446972A (en) * | 2018-10-24 | 2019-03-08 | 电子科技大学中山学院 | Gait recognition model establishing method, recognition method and device based on electromyographic signals |
| CN109446972B (en) * | 2018-10-24 | 2021-08-31 | 电子科技大学中山学院 | Gait recognition model establishing method, recognition method and device based on electromyographic signals |
| CN109512390A (en) * | 2018-10-26 | 2019-03-26 | 武汉大学 | Sleep stage method and wearable device based on EEG time domain various dimensions feature and M-WSVM |
| CN109498370A (en) * | 2018-12-26 | 2019-03-22 | 杭州电子科技大学 | Joint of lower extremity angle prediction technique based on myoelectricity small echo correlation dimension |
| CN109498009A (en) * | 2018-12-26 | 2019-03-22 | 杭州电子科技大学 | Human body daily behavior myoelectricity feature selection approach based on feature class separability index |
| CN110414619A (en) * | 2019-08-05 | 2019-11-05 | 重庆工商职业学院 | One kind being based on EMG signal recognition methods |
| CN110414619B (en) * | 2019-08-05 | 2023-04-07 | 重庆工商职业学院 | EMG signal identification method |
| CN110672090A (en) * | 2019-09-26 | 2020-01-10 | 深圳大学 | Indoor and outdoor integrated positioning method and system based on electromyographic data |
| CN114344077A (en) * | 2021-12-07 | 2022-04-15 | 华南理工大学 | Flexible upper limb rehabilitation robot system based on SEMG movement intention recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107440716A (en) | Human body lower limbs athletic performance classification discrimination method based on single channel electromyographic signal | |
| CN101690659B (en) | Brain wave analysis method | |
| Roy et al. | A combined sEMG and accelerometer system for monitoring functional activity in stroke | |
| Cifrek et al. | Surface EMG based muscle fatigue evaluation in biomechanics | |
| Li et al. | Teager–Kaiser energy operation of surface EMG improves muscle activity onset detection | |
| CN105877766A (en) | Mental state detection system and method based on multiple physiological signal fusion | |
| CN110353702A (en) | A kind of emotion identification method and system based on shallow-layer convolutional neural networks | |
| CN204931634U (en) | Based on the depression evaluating system of physiologic information | |
| CN109745033A (en) | Dynamic electrocardiogram method for evaluating quality based on time-frequency two-dimensional image and machine learning | |
| CN105147248A (en) | Physiological information-based depressive disorder evaluation system and evaluation method thereof | |
| CN102184415B (en) | Electroencephalographic-signal-based fatigue state recognizing method | |
| CN103584872A (en) | Psychological stress assessment method based on multi-physiological-parameter integration | |
| CN105997064A (en) | Method for identifying human lower limb surface EMG signals (electromyographic signals) | |
| CN110321783A (en) | A kind of MEG spike detection method and system based on 1D convolutional neural networks | |
| CN103610447A (en) | Mental workload online detection method based on forehead electroencephalogram signals | |
| CN110123342B (en) | Internet addiction detection method and system based on brain waves | |
| CN105942974A (en) | Sleep analysis method and system based on low frequency electroencephalogram | |
| CN104571504B (en) | A kind of online brain-machine interface method based on Imaginary Movement | |
| CN112137616B (en) | Consciousness detection device for multi-sense brain-body combined stimulation | |
| CN105342605A (en) | Method for removing myoelectricity artifacts from brain electrical signals | |
| CN105147282A (en) | Cognitive disorder electroencephalogram recognition system | |
| CN109480872A (en) | Driving fatigue detection method based on EEG signals frequency band energy than feature | |
| CN105942975A (en) | Stable state visual sense induced EEG signal processing method | |
| CN110537913A (en) | LDA algorithm-based channel selection method for surface electromyographic signals | |
| Ahmed et al. | EMG Signal classification for detecting neuromuscular disorders |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171208 |
|
| WD01 | Invention patent application deemed withdrawn after publication |