CN102722643A - Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb - Google Patents
Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb Download PDFInfo
- Publication number
- CN102722643A CN102722643A CN2012101688147A CN201210168814A CN102722643A CN 102722643 A CN102722643 A CN 102722643A CN 2012101688147 A CN2012101688147 A CN 2012101688147A CN 201210168814 A CN201210168814 A CN 201210168814A CN 102722643 A CN102722643 A CN 102722643A
- Authority
- CN
- China
- Prior art keywords
- signal
- control
- sigma
- function
- value
- 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
Images
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to medical rehabilitation instruments. In order to efficiently improve the control precision and the speed of the control of an artificial limb, the invention adopts the technical scheme that the multi-information merging movement mode identification method used for the control of an artificial lower surface electromyogram limb comprises the following steps: extracting electromyogram signals of muscles when people do sports, calculating singular values and a spectral entropy of the electromyogram signals to be used as characteristic parameters of the mode recognition; constructing an action recognition model by a support vector machine, and therefore, the four common action modes, namely squatting, standing, extending the knee joint, and walking can be entirely and accurately distinguished. The multi-information merging movement mode identification method is mainly applied in the design and the manufacture of the medical rehabilitation instruments.
Description
Technical field
The present invention relates to the medical rehabilitation apparatus, specifically, relate to the many information fusion motor pattern recognition methods that is used for the control of lower limb myoelectric limb.
Background technology
Along with the development of the deep of electromyographic signal physiological Study and checking with EMG method technology, the application of surface electromyogram signal (sEMG) more and more widely comprises medical science of recovery therapy, artificial limb and computer control etc., the particularly control of disabled person's myoelectric limb.Through lay surface electromyogram signal (the surface electromyogram signal that electrode detection arrives at the body surface certain position; SEMG); And the relevance of various degrees between the functional status of muscle and the active state can reflect nervimuscular activity to a certain extent., nothing portable because of it created, safe, many target spot is measured, and bionical control is effective, has a extensive future, and is widely used in the clinical rehabilitation process.Along with science and technology development; The research of myoelectric-controlled prosthesis has had significant progress; But be not used widely yet; It is main that present myoelectric limb is still done evil through another person with upper limbs, and is mostly doing evil through another person of a better simply 2-3 degree of freedom, person's reset terminal water tumbler that only can make the deformed limb, take knife and fork, simple motor function such as hold a pen.Also do not see technology maturation on the market, control the birth of myoelectricity artificial leg accurately; Its main cause is that the processing of surface electromyogram signal and the identification of action mode still have many problems to need to be resolved hurrily; Set up the neural flesh bone model of human body lower limbs, the research of carrying out lower limb myoelectricity and kinematics aspect has very important meaning.
In a mode identification procedure, no matter be computing machine or people, at first all to find out some the most representative characteristics earlier, could go identification according to these characteristics then.Electromyographic signal (EMG) is a kind of very faint bioelectrical signals; Want artificial limb is realized accurately control; Precondition is feature extraction accurately and reliably and effective action recognition; Therefore the electromyographic signal characteristic has crucial meaning to motion analysis, and traditional myoelectricity feature extracting method generally has time-domain analysis method, frequency-domain analysis method, Time-Frequency Analysis Method and Nonlinear Dynamics.The characteristic identifying parameter that merges time-frequency characteristic has been extracted in this research, carries out Classification and Identification, and is reliable and have an actual using value.
Summary of the invention
The present invention is intended to overcome the deficiency of prior art; Improve the accurate control and the speed of artificial limb control effectively; For achieving the above object, the technical scheme that the present invention takes is to be used for many information fusion motor pattern recognition methods of lower limb myoelectric limb control; Comprise the following steps: through extracting the electromyographic signal of human body muscle when moving; Calculate electromyographic signal singular value and Power Spectral Entropy as the characteristic parameter of pattern-recognition, utilize SVMs SVM to set up the action recognition model, thereby comprehensive and accurate differentiation is squatted down, stands, stretched knee and walks four kinds of common action mode.
The electromyographic signal singular value be matrix intrinsic characteristic, the definition of Singular Value Decomposition Using is following:
Suppose that M is the matrix of k*e, k>e, order is r, r≤e then exists e*e orthogonal matrix Z and k*k orthogonal matrix Q, makes:
Q
TMZ=∑
∑ is the non-negative diagonal matrix of k*e:
S=diag (ε
1, ε
2, ε
3... ε
r)
Wherein, diag representes diagonal matrix, ε
1, ε
2, ε
3... ε
rTogether with ε
R+1=ε
R+2=... ε
e=0 is called the singular value of M, and the column vector of Z, Q is respectively the left and right singular vector of M;
Power Spectral Entropy defines as follows: use U={u
1, u
2, u
3..., u
n, certain uncertain system is represented in n>=1, wherein, each value Making by Probability Sets representes with P,
p={p
1,p
2,p
3,......,p
n},0≤p
i≤1,i=1,2,3.....,n
And
So, the information entropy of system is expressed as:
The entropy of the power spectrum that gets after the signal process FFT conversion just is called Power Spectral Entropy; Electromyographic signal Power Spectral Entropy computing method are following:
1) signal is carried out the FFT conversion, obtain its DFT X (w
i), w
iBe i Frequency point;
2) calculating its power spectrum density is:
Wherein, X (w
i) be the Fourier transform of i Frequency point, N is the number a little of getting, p (w
i) be power spectrum density;
3) the power spectrum density distribution function p of signal calculated next
i, i.e. value probability:
4) Power Spectral Entropy of signal calculated:
Wherein, p
iBeing the 3rd) calculate i the power spectrum density distribution function of putting is the value probability in the step.
Utilize the svm classifier process following:
Consider the linear regression problem of n training sample, establish training dataset (x
i, y
i), i=1 ..., l is the input pattern of i sample, y
iCorresponding to the desired output of i sample, at first with the signal value X=(x of a nonlinear transformation with input
1, x
2..., x
1) be transformed in certain higher dimensional space, ask the optimal classification face at transformation space then, this conversion only needs an inner product function (x
iY
j), i, j=1 ..., l, x
i, y
j∈ X, X is the signal value of input; Select for use the radially basic kernel function RBF of Gauss to change, its expression formula is following:
K(x
i,x
j)=exp{-γ|x
i-x
j|
2}
Wherein, K (x
i, x
j) be selected kernel function, γ is a nuclear parameter, need seek optimum value with the mode of cross validation, like this, former sample space just has been mapped in the high-dimensional feature space, constructs optimal decision function in this high-dimensional feature space:
y=w×K(x
i,x
j)+b
Y is output, x
i, x
j∈ X is an input signal, and w, b are required parameter; Like this, the nonlinear estimation function is converted into the Linear Estimation function of high-dimensional feature space; Definition R:
In the formula, || w||
2Be the complexity of controlling models, c is a regularization parameter, and control is to exceeding the punishment degree of error sample;
Be the error control function; Utilize structural risk minimization, seek w, b minimizes R exactly; Therefore, classification function is expressed as:
In the formula, y is output, α
iBe Lagrangian coefficient; B is the basic parameter of lineoid, and the characteristic parameter of input electromyographic signal is output as type of action; Set up model and inspection-classification result.
Technical characterstic of the present invention and effect:
Purport of the present invention is to propose a kind of many information fusion motor pattern identification new method that is used for the control of lower limb myoelectric limb; Set up model of cognition through SVMs; Contrast the classification results under single feature mode and the many information fusion pattern situation, search out the most rapidly, myoelectric limb control method the most accurately.This invention can obtain considerable social benefit and economic benefit.Optimum implementation is intended and is adopted patent transfer, technological cooperation or product development.
Description of drawings
Fig. 1 is used for many information fusion motor pattern identification new method system chart of lower limb myoelectric limb control.
Fig. 2 list feature mode and Feature Fusion pattern-recognition result contrast.
Embodiment
The present invention provides a kind of many information fusion motor pattern identification new method that is used for the control of lower limb myoelectric limb; Through extracting the electromyographic signal of human body muscle when moving; Calculate its singular value and Power Spectral Entropy characteristic parameter as pattern-recognition; Utilize SVMs (SVM) method to set up the action recognition model, thereby comprehensive and accurate differentiation is squatted down, stand, is stretched knee and walk four kinds of common action mode.This method is a kind of brand-new lower limb movement recognition technology, belongs to disability rehabilitation's technical field of medical instruments.
Purport of the present invention is through extracting the myoelectric information of human body lower limbs related muscles when doing different action, calculate its various features parameter, in the input svm classifier recognizer, carrying out modeling and action recognition.This invention can improve the accurate control and the speed of artificial limb control effectively, and obtains considerable social benefit and economic benefit.
As shown in the figure; The present invention utilizes the Noraxon surface myoelectric wireless telemetering acquisition analysis system TeleMyo2400DTS of produced in usa; Extract human body lower limbs rectus femoris, biceps muscle of thigh, semitendinosus, musculus soleus, gastrocnemius and tibialis anterior squat down, stand, stretch knee and four patterns of walking under pre-activity make myoelectric information; Through preprocessor, myoelectricity is carried out amplification filtering, extract singular value and Power Spectral Entropy Feature Fusion parameter as pattern identification; Utilize the method for support vector base to set up disaggregated model, thereby realize the accurately purpose of identification.
The characteristic parameter of 1 extraction
Has bigger class spacing in order both to satisfy characteristic parameter; Have lower computation complexity again, this paper extracts AR model parameter, cepstrum coefficient, singular value and the Power Spectral Entropy characteristic parameter as the myoelectricity action recognition, the input category device; Move the identification of mode, the concrete introduction as follows.
1.1 singular value
Singular value be matrix intrinsic characteristic, meet desired stability of electromyographic signal feature identification and rotation, constant rate property, be an effective patterns characteristic, can effectively reflect the myoelectricity frequency domain character.The definition of Singular Value Decomposition Using is following:
Suppose that M is the matrix of k*e, k>e, order is r, r≤e then exists e*e orthogonal matrix Z and k*k orthogonal matrix Q, makes:
Q
TMZ=∑
∑ is the non-negative diagonal matrix of k*e:
S=diag (ε
1, ε
2, ε
3... ε
r)
Wherein, diag representes diagonal matrix, ε
1, ε
2, ε
3... ε
rTogether with ε
R+1=ε
R+2=... ε
e=0 is called the singular value of M, and the column vector of Z, Q is respectively the left and right singular vector of M.
1.2 Power Spectral Entropy
Entropy is as a non-linear dynamic mathematic(al) parameter, can be used for the confusion degree of descriptive system, the incidence of fresh information in the measurement system; The structure situation of the spectrum of reflection time series signal; Signal energy distributes even more at each sub-band, then entropy is big more, and expression signal is more complicated; Uncertainty is big more, can be used as the tolerance of system complexity.Therefore, select the characteristic parameter of Power Spectral Entropy, can not only distinguish electromyographic signal and noise signal effectively, have good noise robustness simultaneously, disclose the energy distribution characteristic in the time frequency space as the myoelectricity action recognition.
Power Spectral Entropy defines as follows: use U={u
1, u
2, u
3..., u
n, certain uncertain system is represented in n>=1, wherein, each value Making by Probability Sets representes with P,
p={p
1,p
2,p
3,......,p
n},0≤p
i≤1,i=1,2,3.....,n
And
So, the information entropy of system is expressed as:
It is Power Spectral Entropy that the entropy of the power spectrum that gets after the signal process FFT conversion just is called.Its computing method of the Power Spectral Entropy of myoelectricity are following among the present invention:
1) input signal is carried out the FFT conversion, obtain its DFT X (w
i), w
iBe i Frequency point.
2) calculating its power spectrum density is:
Wherein, X (w
i) be the Fourier transform of i Frequency point, N is the number a little of getting, p (w
i) be power spectrum density;
3) the power spectrum density distribution function p of signal calculated next
i, i.e. value probability:
4) Power Spectral Entropy of signal calculated:
Wherein, p
iBe i the power spectrum density distribution function of putting, i.e. the value probability that calculated in the 3rd step.
2 Classification and Identification principles
SVMs (Support Vector Machine; SVM) be a kind of new machine learning techniques that develops with the research group of its leader's AT&T Labs by Vapnik; Because its outstanding learning performance, this technology has become the research focus of current international machine learning circle.The theoretical foundation of SVMs is VC dimension theory and structural risk minimization, and wherein, the VC dimension refers to the complexity of problem, is to make up model to get approximation ratio with actual value and structure risk is described, by empiric risk with put that trade wind is dangerous to be formed.Empiric risk is meant the identification error of sorter on sample data, puts the popularization degree that the trade wind danger has then characterized sorter, represents us can to what extent trust this sorter, and it comes common decision by sample size and VC dimension.Obviously, the VC dimension is high more, and it is just poor more that sorter is promoted performance.SVM just is based on a kind of sorting algorithm on the two theoretical foundation.
Consider the linear regression problem of n training sample, establish training dataset (x
i, y
i), i=1 ..., l is the input pattern of i sample, y
iCorresponding to the desired output of i sample, at first use a nonlinear transformation with input variable X=(x
1, x
2..., x
1) be transformed in certain higher dimensional space, ask the optimal classification face at transformation space then.This conversion only needs an inner product function (x
iY
j), i, j=1 ..., l, x
i, y
j∈ X, X is the signal value of input.This paper selects for use the radially basic kernel function of Gauss (RBF) to change, and its expression formula is following:
K(x
i,x
j)=exp{-γ|x
i-x
j|
2}
Wherein, K (x
i, x
j) be selected kernel function, x
i, x
j∈ X is an input signal values, and γ is a nuclear parameter, need seek optimum value with the mode of cross validation.Like this, former sample space just has been mapped in the high-dimensional feature space, constructs optimal decision function in this high-dimensional feature space:
y=w×K(x
i,x
j)+b
Y is output, x
i, x
j∈ X is an input signal values, and w, b are required parameter.Like this, the nonlinear estimation function is converted into the Linear Estimation function of high-dimensional feature space.Definition R:
In the formula, || w||
2Be the complexity of controlling models, c is a regularization parameter, and control is to exceeding the punishment degree of error sample;
Be the error control function; Utilize structural risk minimization, seek w, b minimizes R exactly.Therefore, classification function can be expressed as among the present invention:
In the formula, y is output, α
iBe Lagrangian coefficient, γ is a nuclear parameter, and b is the basic parameter of lineoid.Import the characteristic parameter of six tunnel electromyographic signals, be output as type of action.Set up model and inspection-classification result.
Beneficial effect
Table 1 and table 2 are respectively 10 and are used separately singular values and Power Spectral Entropy by examination, the recognition result that carries out the classification of motion, and table 3 is for carrying out singular value and Power Spectral Entropy the recognition result after the Feature Fusion.
Table 1 singular value is as the correct recognition rata of proper vector
Table 2 Power Spectral Entropy is as the correct recognition rata of proper vector
Table 3 singular value+Power Spectral Entropy is as the correct recognition rata of proper vector
As can beappreciated from fig. 2, the perfect adaptation of different value and Power Spectral Entropy makes the pattern-recognition result all remain on about 90% basically; Average recognition correct rate is 91.3%; The highest discrimination reaches 97.16%, far above the pattern-recognition of single characteristic, is ideal characteristic parameter; Can be used for the accurate control of lower limb myoelectric limb, huge potential using value is arranged.
Purport of the present invention is to propose a kind of many information fusion motor pattern identification new method that is used for the control of lower limb myoelectric limb; Set up model of cognition through SVMs; Contrast the classification results under single feature mode and the many information fusion pattern situation, search out the most rapidly, myoelectric limb control method the most accurately.This invention can obtain considerable social benefit and economic benefit.Optimum implementation is intended and is adopted patent transfer, technological cooperation or product development.
Claims (3)
1. many information fusion motor pattern recognition methods that is used for the control of lower limb myoelectric limb; It is characterized in that; Comprise the following steps: through extracting the electromyographic signal of human body muscle when moving; Calculate electromyographic signal singular value and Power Spectral Entropy as the characteristic parameter of pattern-recognition, utilize SVMs SVM to set up the action recognition model, thereby comprehensive and accurate differentiation is squatted down, stands, stretched knee and walks four kinds of common action mode.
2. the many information fusion motor pattern recognition methods that is used for lower limb myoelectric limb control as claimed in claim 1 is characterized in that, the electromyographic signal singular value be matrix intrinsic characteristic, the definition of Singular Value Decomposition Using is following:
Suppose that M is the matrix of k*e, k>e, order is r, r≤e then exists e*e orthogonal matrix Z and k*k orthogonal matrix Q, makes:
Q
TMZ=∑
∑ is the non-negative diagonal matrix of k*e:
S=diag (ε
1, ε
2, ε
3... ε
r)
Wherein, diag representes diagonal matrix, ε
1, ε
2, ε
3... ε
rTogether with ε
R+1=ε
R+2=... ε
e=0 is called the singular value of M, and the column vector of Z, Q is respectively the left and right singular vector of M;
Power Spectral Entropy defines as follows: use U={u
1, u
2, u
3..., u
n, certain uncertain system is represented in n>=1, wherein, each value Making by Probability Sets representes with P,
p={p
1,p
2,p
3,......,p
n},0≤p
i≤1,i=1,2,3......,n
And
So, the information entropy of system is expressed as:
The entropy of the power spectrum that gets after the signal process FFT conversion just is called Power Spectral Entropy; Electromyographic signal Power Spectral Entropy computing method are following:
1) signal is carried out the FFT conversion, obtain its DFT X (w
i), w
iBe i Frequency point;
2) calculating its power spectrum density is:
Wherein, X (w
i) be the Fourier transform of i Frequency point, N is the number a little of getting, p (w
i) be power spectrum density;
3) the power spectrum density distribution function p of signal calculated next
i, i.e. value probability:
4) Power Spectral Entropy of signal calculated:
Wherein, p
iBeing the 3rd) calculate i the power spectrum density distribution function of putting is the value probability in the step.
3. the many information fusion motor pattern recognition methods that is used for the control of lower limb myoelectric limb as claimed in claim 1 is characterized in that, utilizes the svm classifier process following:
Consider the linear regression problem of n training sample, establish training dataset (x
i, y
i), i=1 ..., 1, be the input pattern of i sample, y
iCorresponding to the desired output of i sample, at first with the signal value X=(x of a nonlinear transformation with input
1, x
2..., x
1) be transformed in certain higher dimensional space, ask the optimal classification face at transformation space then, this conversion only needs an inner product function (x
iY
j), i, j=1 ..., l, x
i, y
j∈ X, X is the signal value of input; Select for use the radially basic kernel function RBF of Gauss to change, its expression formula is following:
K(x
i,x
j)=exp{-γ|x
i-x
j?|
2}
Wherein, K (x
i, x
j) be selected kernel function, γ is a nuclear parameter, need seek optimum value with the mode of cross validation, like this, former sample space just has been mapped in the high-dimensional feature space, constructs optimal decision function in this high-dimensional feature space:
y=w×K(x
i,x
j)+b
Y is output, x
i, x
j∈ X is an input signal values, and w, b are required parameter; Like this, the nonlinear estimation function is converted into the Linear Estimation function of high-dimensional feature space; Definition R:
In the formula, || w||
2Be the complexity of controlling models, c is a regularization parameter, and control is to exceeding the punishment degree of error sample;
Be the error control function; Utilize structural risk minimization, seek w, b minimizes R exactly; Therefore, classification function is expressed as:
In the formula, y is output, α
iBe Lagrangian coefficient; B is the basic parameter of lineoid, and the characteristic parameter of input electromyographic signal is output as type of action; Set up model and inspection-classification result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012101688147A CN102722643A (en) | 2012-05-28 | 2012-05-28 | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012101688147A CN102722643A (en) | 2012-05-28 | 2012-05-28 | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102722643A true CN102722643A (en) | 2012-10-10 |
Family
ID=46948400
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2012101688147A Pending CN102722643A (en) | 2012-05-28 | 2012-05-28 | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102722643A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102961203A (en) * | 2012-12-10 | 2013-03-13 | 杭州电子科技大学 | Method for identifying surface electromyography (sEMG) on basis of empirical mode decomposition (EMD) sample entropy |
CN103908361A (en) * | 2014-04-02 | 2014-07-09 | 韩晓新 | Method for acquiring and operating artificial limb joint movement coupling drive signals |
CN105139038A (en) * | 2015-09-14 | 2015-12-09 | 李玮琛 | Electromyographic signal mode identification method |
CN105326501A (en) * | 2015-12-10 | 2016-02-17 | 宁波工程学院 | Muscle disease monitoring method based on sEMG |
CN107753044A (en) * | 2017-10-19 | 2018-03-06 | 泰山医学院 | A kind of ankle-joint varus inducing device and the method for measurement musculus peroneus longus reaction |
CN107832686A (en) * | 2017-10-26 | 2018-03-23 | 杭州电子科技大学 | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal |
CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人(合肥)国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
CN109846481A (en) * | 2018-12-25 | 2019-06-07 | 北京津发科技股份有限公司 | Surface electromyogram signal treating method and apparatus, data processing equipment and storage medium |
CN109864740A (en) * | 2018-12-25 | 2019-06-11 | 北京津发科技股份有限公司 | A kind of the surface electromyogram signal acquisition sensor and equipment of motion state |
CN112036357A (en) * | 2020-09-09 | 2020-12-04 | 曲阜师范大学 | Upper limb action recognition method and system based on surface electromyogram signal |
CN117038006A (en) * | 2023-07-21 | 2023-11-10 | 筋斗云易行科技(西安)有限责任公司 | Method for performing rehabilitation training AI auxiliary diagnosis decision after upper and lower limb orthopedics operation |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833671A (en) * | 2010-03-30 | 2010-09-15 | 西安理工大学 | Support vector machine-based surface electromyogram signal multi-class pattern recognition method |
-
2012
- 2012-05-28 CN CN2012101688147A patent/CN102722643A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833671A (en) * | 2010-03-30 | 2010-09-15 | 西安理工大学 | Support vector machine-based surface electromyogram signal multi-class pattern recognition method |
Non-Patent Citations (2)
Title |
---|
GABRIELLA BALESTRA ET AL.: "Time-Frequency Analysis of Myoelectric Signals During Dynamic Comtractions", 《ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE,IEEE》 * |
任亚莉: "基于功率谱熵和频带能量的运动意思任务分类研究", 《计算机应用与软件》 * |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102961203B (en) * | 2012-12-10 | 2015-04-22 | 杭州电子科技大学 | Method for identifying surface electromyography (sEMG) on basis of empirical mode decomposition (EMD) sample entropy |
CN102961203A (en) * | 2012-12-10 | 2013-03-13 | 杭州电子科技大学 | Method for identifying surface electromyography (sEMG) on basis of empirical mode decomposition (EMD) sample entropy |
CN103908361A (en) * | 2014-04-02 | 2014-07-09 | 韩晓新 | Method for acquiring and operating artificial limb joint movement coupling drive signals |
CN105139038A (en) * | 2015-09-14 | 2015-12-09 | 李玮琛 | Electromyographic signal mode identification method |
CN105326501A (en) * | 2015-12-10 | 2016-02-17 | 宁波工程学院 | Muscle disease monitoring method based on sEMG |
CN107753044B (en) * | 2017-10-19 | 2021-12-31 | 山东第一医科大学(山东省医学科学院) | Ankle joint varus inducing device and method for measuring fibula longus muscle reaction |
CN107753044A (en) * | 2017-10-19 | 2018-03-06 | 泰山医学院 | A kind of ankle-joint varus inducing device and the method for measurement musculus peroneus longus reaction |
CN107832686A (en) * | 2017-10-26 | 2018-03-23 | 杭州电子科技大学 | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal |
CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人(合肥)国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
CN109846481A (en) * | 2018-12-25 | 2019-06-07 | 北京津发科技股份有限公司 | Surface electromyogram signal treating method and apparatus, data processing equipment and storage medium |
CN109864740A (en) * | 2018-12-25 | 2019-06-11 | 北京津发科技股份有限公司 | A kind of the surface electromyogram signal acquisition sensor and equipment of motion state |
CN112036357A (en) * | 2020-09-09 | 2020-12-04 | 曲阜师范大学 | Upper limb action recognition method and system based on surface electromyogram signal |
CN112036357B (en) * | 2020-09-09 | 2023-05-12 | 曲阜师范大学 | Upper limb action recognition method and system based on surface electromyographic signals |
CN117038006A (en) * | 2023-07-21 | 2023-11-10 | 筋斗云易行科技(西安)有限责任公司 | Method for performing rehabilitation training AI auxiliary diagnosis decision after upper and lower limb orthopedics operation |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102722643A (en) | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb | |
Lu et al. | A study of personal recognition method based on EMG signal | |
CN102961203B (en) | Method for identifying surface electromyography (sEMG) on basis of empirical mode decomposition (EMD) sample entropy | |
CN105877766B (en) | A kind of state of mind detection system and method based on the fusion of more physiological signals | |
CN102930284B (en) | Surface electromyogram signal pattern recognition method based on empirical mode decomposition and fractal | |
He et al. | Invariant surface EMG feature against varying contraction level for myoelectric control based on muscle coordination | |
CN101317794B (en) | Myoelectric control ability detecting and training method for hand-prosthesis with multiple fingers and multiple degrees of freedom | |
CN104127181B (en) | Electromyographic signal tumble detection method based on WKFDA | |
CN104107042A (en) | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine | |
CN107832686A (en) | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal | |
CN102799937A (en) | Lower limb movement track predication method under fusion of information of myoelectricity signal and joint angle | |
CN101732110B (en) | Method for fusing brain electricity and muscle electricity signal chaos characteristics for hand motion identification | |
CN105446484A (en) | Electromyographic signal gesture recognition method based on hidden markov model | |
Shao et al. | Single-channel SEMG using wavelet deep belief networks for upper limb motion recognition | |
Mokhlesabadifarahani et al. | EMG signals characterization in three states of contraction by fuzzy network and feature extraction | |
Oweis et al. | ANN-based EMG classification for myoelectric control | |
CN109498370A (en) | Joint of lower extremity angle prediction technique based on myoelectricity small echo correlation dimension | |
Su et al. | Measurement of upper limb muscle fatigue using deep belief networks | |
CN106845348A (en) | A kind of gesture identification method based on arm surface electromyographic signal | |
Veer | A flexible approach for segregating physiological signals | |
Kurzynski et al. | Multiple classifier system applied to the control of bioprosthetic hand based on recognition of multimodal biosignals | |
CN104850735A (en) | Activity recognition method based on stack own coding | |
CN109766843A (en) | EMG Feature Extraction based on improved limited Boltzmann machine | |
Veer et al. | Analysis and recognition of operations using SEMG from upper arm muscles | |
CN114569143A (en) | Myoelectric gesture recognition method based on attention mechanism and multi-feature fusion |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C05 | Deemed withdrawal (patent law before 1993) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20121010 |