CN107832686A - Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal - Google Patents
Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal Download PDFInfo
- Publication number
- CN107832686A CN107832686A CN201711013730.5A CN201711013730A CN107832686A CN 107832686 A CN107832686 A CN 107832686A CN 201711013730 A CN201711013730 A CN 201711013730A CN 107832686 A CN107832686 A CN 107832686A
- Authority
- CN
- China
- Prior art keywords
- signal
- entropy
- class
- acceleration
- euclidean distance
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention proposes the lower limb motion mode recognition methods of fusion surface myoelectric and acceleration signal.First, human body lower limbs surface electromyogram signal, acceleration signal are obtained;Surface electromyogram signal is decomposed into multiple multiplicative functions with local mean value decomposition algorithm, according to the average Euclidean distance for characterizing different action separation properties, determine local mean value decomposition algorithm decompose after first multiplicative function multiple dimensioned arrangement entropy, extract first multiplicative function multiple dimensioned arrangement entropy as surface electromyogram signal feature.The importance of different scale entropy is calculated, determines that Scale Entropy forms 4 dimensional feature vectors, and 7 dimensional feature vectors are formed with the ordering entropy of 3-axis acceleration;By 7 dimensional feature vectors input improved binary-tree support vector machine progress lower limb motion mode identification according to average sample distribution between Euclidean distance and class in class.The achievable human body lower limbs motion intention of the present invention is real-time, accurately identifies, and recognition result can be used for exoskeleton robot interactive controlling etc..
Description
Technical field
The invention belongs to area of pattern recognition, is related to a kind of fusion surface electromyogram signal (sEMG), the people of acceleration signal
Body lower limb motion mode recognition methods, example to walking, upstairs, downstairs, stand to sit, sit on station, stand to crouching and crouching arrive at a station this seven
Daily behavior action is identified, and obtains preferable recognition effect.
Background technology
With the fast development of society, influenceed to trigger lower limb one side extremity motor function damage by congenital environment and acquired disease
The patient of wound is increasing.Improve the quality of life of this kind of patient and it is recovered locomitivity and have become social concerns
Focus and medical rehabilitation field theme.But due to medical institutions, technician and equipment shortage and expensive expensive big limitation
The development of rehabilitation, and traditional rehabilitation maneuver doctor's working strength is big, and patient's property of participation is low, training effect and evaluation
Influenceed by doctor's subjective consciousness, therapeutic effect is extremely limited.With the rapid development of robot technology, generate with reference to ectoskeleton
The advanced rehabilitation maneuver such as healing robot, interactive information between extract real-time human body and exoskeleton robot, establishes ectoskeleton
The active flexible control strategy of robot system, it is possible to achieve the initiative rehabilitation being intended to according to human motion is trained.And under human body
Limb motion intention is in real time, to accurately identify be the key for realizing exoskeleton robot interactive controlling, therefore herein to lower extremity movement mould
Accurately identifying for formula intention has made intensive studies.
Surface electromyogram signal (Surface electromyography, sEMG) is widely used in motion intention identification neck
Domain, but due in electromyographic signal uneven stability caused by noise and human biological signal's noise influence, right
SEMG must carry out denoising Processing before further handling.Empirical mode decomposition (Empirical mode decomposition,
EMD) it can make sophisticated signal be decomposed into limited individual intrinsic mode function (Intrinsic mode function, IMF), divide
Each IMF components of solution out contain the sample characteristics of the different time scales of primary signal, are to be based on signal sequence due to decomposing
The local characteristicses of row time scale, therefore there is adaptivity, relative to Fu in short-term on processing non-stationary and nonlinear properties
In the Time-Frequency Analysis method such as leaf transformation, wavelet transformation have obviously advantage, once proposing just in different engineering fields
It is widely used, but because EMD has boundary effect and modal overlap, a kind of new adaptive Time Frequency Analysis method is local
Average decomposes (Local mean decomposition, LMD) can be by a complicated non-stationary multi -components amplitude-frequency modulated signal
Decompose several PF (Product functions, PF) components with physical significance.
After LMD decomposition, major problem is that how from the PF constituents extraction motion characteristic information obtained.When common
Characteristic of field (such as absolute mean, root mean square, waveform length, zero passage points), frequency domain character (such as autoregression model coefficient, average frequency
Rate, median frequency), nonlinear analysis method (such as Hilbert-Huang transform), Liapunov exponent, Sample Entropy, approximate entropy
SEMG sign extraction field is widely used to, but the above method is present that anti-noise ability is weak, process of data preprocessing is complicated etc.
Limitation, causes the minutias of sEMG to be in itself difficult to effectively extract.Ordering entropy (Permutation entropy, PE) is used as one
Nonlinear kinetics parameter of the kind based on complexity measure, has the advantages that calculating speed is fast, anti-noise ability is strong, gradually applies
In the analysis of the complex biological electric signals such as electrocardio, brain electricity.But above time domain and Time-Frequency Analysis method are confined to the single time
Yardstick, and time-domain signal is handled and is decomposed into different time scales by multiscale analysis method by multiple dimensioned coarse, it is multiple dimensioned
Analysis bag more detailed information on different time scales containing primary signal, electromyographic signal can be more reacted under different motion pattern
Complicated bulk properties.
Acceleration signal can not only reflect daily behavior action caused velocity information and in space over time
Motion track information, can also provide daily behavior action relative to acceleration of gravity direction angle tilt information, therefore
It is widely used in daily behavior motion detection and fall detection, the normal temporal signatures such as simple threshold values for extracting acceleration signal,
Kurtosis, signal amplitude vector, signal amplitude domain (Signal Magnitude Area, SMA) and acceleration transient change value etc.,
Irene S etc. using acceleration to sitting, standing, walking is identified, obtain 99% recognition correct rate.But believed using acceleration
Number simple temporal signatures can not effectively reflect the difference between lower limb different motion pattern, and what ordering entropy can prepare
Reflect the complexity of different motion pattern acceleration signal, favorably distinguish different patterns.
The content of the invention
In order to realize lower limb multi-pattern recognition, the present invention proposes a kind of fusion surface myoelectric and acceleration signal
Lower limb motion mode recognition methods.First, it is multiple multiplicative functions (PF) to be decomposed with local mean value by Decomposition Surface EMG,
The multiple dimensioned ordering entropy of PF compositions is calculated again.Then, it is selected per road myoelectricity by Laplce's weight (LS) feature selecting algorithm
One yardstick ordering entropy of signal is characterized, and the ordering entropy composition characteristic of this feature and acceleration signal vector.Finally, root
According to sample distribution between Euclidean distance and class in class, it is proposed that improved binary-tree support vector machine (ISVM-BT), characteristic vector
Input the SVMs and realize that lower limb motion mode is classified.
In order to realize the above object the inventive method mainly includes the following steps that:
Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal, it is characterised in that:This method includes
Following steps:
Step 1. obtains human body lower limbs surface electromyogram signal, acceleration signal;Detailed process is as follows:
Four pieces of selected tibialis anterior, gastrocnemius, rectus femoris, semitendinosus muscle pass through myoelectricity acquisition system as signal source
Obtain electromyographic signal;One 3-axis acceleration sensor node is fixed on tester's waist, captures and records the axle of x, y, z three
Original acceleration exercise data.
The surface electromyogram signal that step 1 obtains is decomposed into multiple multiplicative functions by step 2. with local mean value decomposition algorithm;
Step 3. determines that local mean value decomposition algorithm decomposes it according to the average Euclidean distance for characterizing different action separation properties
The multiple dimensioned arrangement entropy of first multiplicative function afterwards, the multiple dimensioned arrangement entropy for extracting first multiplicative function are believed as surface myoelectric
Number feature;
Step 4. application Laplce's weight feature selecting algorithm calculates the weight of different scale entropy, by Laplce's weight
It is worth four Scale Entropies of highest and forms four dimensional feature vectors;
The ordering entropy of four dimensional feature vectors that step 5. is obtained step 4 and 3-axis acceleration form 7 degree of freedom feature to
Amount;
Step 6. by 7 degree of freedom characteristic vector input according to average sample distribution between Euclidean distance and class in class and improved two
Fork tree SVMs carries out lower limb motion mode identification;The kernel function of SVMs is used as by the use of RBF;, pass through something lost
Propagation algorithm obtains global optimum the punishment parameter C and nuclear parameter γ of each subclassification supporting vector machine model.;
Described binary-tree support vector machine algorithmic procedure is as follows:
(1) average Euclidean distance AVI, weight coefficient k average between Euclidean distance AV and class in class are obtained by training datan=
2n, -4 < < n < < 4, n is integer;
(2) for given knCalculate Separatory measure Ii,j, Separatory measure can be by average Euclidean distance AV in classi、
AVjThe Euclidean distance AVI between average classi,jI.e. I is obtained with weight coefficient ki,j=AVIi,j+k(AVi+AVj), i, j=1,
2, N, i ≠ j, it is as follows that symmetrical matrix SI is constructed by Separatory measure
(3) what the hierarchy of binary tree was often gone according to matrix SI sorts with value;
(4) weights k is changednAnd repeat step 6- (2) and step 6- (3) generates a series of binary tree hierarchies;
(5) a binary tree hierarchy and weights k are selectedn, establish SVM sub-classifiers;Need to produce for k class problems
K-1 sub-classifier, classifies to test data, calculates the nicety of grading of grader;
(6) n=n+1 repeat steps 6- (5) is made untill n=4;
(7) according to all weights knCorresponding nicety of grading determines optimal binary tree hierarchy.
It is of the invention with existing lower limb motion mode identify compared with, there are following features:
Due to the complexity of human body attitude information, the multi-mode of single signal such as acceleration or surface electromyogram signal is only relied on
The ambiguity that motion identification has higher False Rate, particularly surface electromyogram signal is extremely strong, is only entered by surface electromyogram signal
The identification of pedestrian's body motion intention is by very big dispute, therefore the knowledge for being integrated into pedestrian's body motion intention for passing through source signal
It is not an effective approach.The present invention devises a kind of lower limb motion mode for merging surface electromyogram signal and acceleration signal
Recognition methods, compared to the simple recognition methods by acceleration or electromyographic signal, greatly improve classification accuracy.
Man-machine interaction is realized as information source using sEMG and acceleration signal, core technology is that human motion is intended to real-time
Accurately identify, and due to the complexity of the complexity of human action, electromyographic signal nonlinear and nonstationary and reality scene so that
Establishing stable, practical man-machine interactive system has larger difficulty.The present invention is being extracted surface electromyogram signal and acceleration letter
After number feature, the feature extracted is sent into support vector machine classifier and carries out lower limb motion mode identification.Due to four road tables
Facial muscle electric signal extracts the higher information redundancy of feature vector dimension that multiple dimensioned ordering entropy is formed and causes pattern-recognition accuracy rate
Decline, computation complexity increase, thus we reduce feature space dimension choose several important scale factor composition characteristics to
Amount, is ranked up according to importance to multiple dimensioned ordering entropy present invention introduces Laplce's Weight algorithm, realizes dimension-reduction treatment.Base
In on the construction that the support vector machine classifier classification performance of binary tree depends primarily on binary tree, the present invention combines Euclidean in class
Sample distribution between distance and class, improve binary-tree support vector machine sorting algorithm.
Brief description of the drawings
Fig. 1 is the implementing procedure figure of the present invention;
Tibialis anterior electromyographic signal and LMD decomposition results when Fig. 2 is walks;
The yardstick that Fig. 3 is tibialis anterior electromyographic signal PF1 arranges entropy distribution;
The yardstick that Fig. 4 is tibialis anterior electromyographic signal PF2 arranges entropy distribution;
Fig. 5 is ISVM-BT algorithm flow charts;
Fig. 6 is the optimal hierarchies of ISVM-BT.
Embodiment
Embodiments of the invention are elaborated below in conjunction with the accompanying drawings:The present embodiment using technical solution of the present invention before
Put and implemented, give detailed embodiment and specific operating process.
As shown in figure 1, the present embodiment comprises the following steps:
Step 1, obtain human body lower limbs surface electromyogram signal, acceleration signal.Detailed process is as follows:
Electromyographic signal is gathered using four-way myoelectricity Acquisition Instrument.Before finally selecting shin bone to leg muscles test and comparison
Flesh, gastrocnemius, rectus femoris, semitendinosus are first used the above-mentioned four pieces of muscle of alcohol wipe before data acquisition is carried out, gone as signal source
Except the soft flocks of skin surface reduces interference, while require that tester not take vigorous exercise in 24h.By an i4Motion tri-
Axle acceleration sensor node is fixed on tester's waist, captures and records the original motion data of the axle of x, y, z three.Due to myoelectricity
The useful energy of signal is mainly distributed between 10Hz~500Hz, for undistorted sampling, therefore sets the sampling of electromyographic signal
Frequency is 1000Hz.Acceleration signal is all in below 20Hz, the present embodiment setting acceleration signal according to caused by physical activity
Sample frequency be 50Hz, data acquisition equipment is transferred data in upper computer software by bluetooth, and host computer is recorded and protected
Deposit the original motion data.The present embodiment mainly have studied away, upstairs, downstairs, stand to sitting, sit on station, stand and arrive at a station this to crouching and crouching
Seven daily behavior actions.Experimental subjects is two health males (24 ± 2 years old, 65 ± 5kg, 170 ± 5cm) and a female
Property (23 years old, 48kg, 162cm).Every group of action repeats 40 times, and each sample is by three axial directions of four tunnel electromyographic signals and x, y, z
Acceleration signal forms, according to the motion frequency of rehabilitation, 2000 follow-up sample point datas of Qu Mei roads electromyographic signal and
100 follow-up sample point datas of acceleration signal are further processed as research object.
The surface electromyogram signal that step (1) obtains is decomposed into multiple multiply by step 2 with local mean value decomposition algorithm (LMD)
Product function (PF).
(LMD) algorithm is decomposed using local mean value to decompose electromyographic signal.The electromyographic signal of tibialis anterior during walking
And decomposition result is as shown in Figure 2.Because the action message of electromyographic signal is concentrated mainly in PF compositions above in therefore figure only
Show preceding 3 PF compositions.
Step 3 determines local mean value decomposition algorithm (LMD) according to the average Euclidean distance for characterizing different action separation properties
The multiple dimensioned arrangement entropy of first multiplicative function (PF1) has higher average Euclidean distance after decomposition, and extraction first multiplies
The multiple dimensioned arrangement entropy of Product function (PF1) is as surface electromyogram signal feature.
LMD decompose after application it is multiple dimensioned arrangement entropy extraction electromyographic signal feature, wherein Embedded dimensions m=5, yardstick because
Sub- τ=20.Fig. 3 and Fig. 4 describes tibialis anterior electromyographic signal PF1 and PF2 yardstick arrangement entropy distribution respectively.
The yardstick arrangement entropy that can be seen that PF1 from Fig. 3 and Fig. 4 has more preferable separation property than PF2, draws to further illustrate
Enter the average Euclidean distance for characterizing different action separation properties
Wherein Tj(τ) is normalized arrangement entropy, and N=7 is amount of action, and M=21 is the number for the Euclidean distance that seven actions are formed
Amount) it is used as evaluation index.The average Euclidean distance of four tunnel electromyographic signals is respectively corresponding to the multiple dimensioned arrangement entropys of PF1 and PF2
1.3169th, 1.2833,1.0483,1.1579 and 0.9607,0.8268,0.7981,0.5050.PF1 compositions have higher be averaged
Euclidean distance this show that PF1 compositions contain more motion characteristic information and have higher separability.Therefore by PF1 into
It is allocated as further being studied for the main component of electromyographic signal.
Step 4 calculates the weighted value of different scale entropy using Laplce's weight feature selecting algorithm, it is determined that (τ=1,
5,4,16) Scale Entropy forms 4 dimensional feature vectors.
If directly the characteristic vector for the dimension of 20 yardsticks arrangement entropy composition 1 × 80 that every road electromyographic signal is calculated is entered
Row calculates, it will increase computation complexity and run time, therefore we using Laplce's weight (Laplacian score,
LS) feature selecting algorithm calculates the importance of different scale entropy.Wherein 20 yardstick ordering entropies of tibialis anterior electromyographic signal are by weight
The property wanted sequence is as follows, and its excess-three road electromyographic signal is similar with this distribution.LS1 < LS7 < LS2 < LS3 < LS4 < LS8 < LS5 <
LS6 < LS9 < LS19 < LS12 < LS10 < LS13 < LS14 < LS11 < LS15 < LS16 < LS17 < LS20 < LS18
The ordering entropy of 4 dimensional feature vectors that step 5 is obtained step (4) and 3-axis acceleration form 7 dimensional features to
Amount.
By experimental analysis and based on final choice from the aspect of computation complexity and accuracy of identification two per road electromyographic signal
The ordering entropy of most important that yardstick (τ=1,5,4,16) entropy and x, y, z 3-axis acceleration forms 1 × 7 dimensional feature vector.
Step 6 by 7 dimensional feature vectors input according to average sample distribution between Euclidean distance and class in class and improved two
Fork tree SVMs (ISVM-BT) carries out lower limb motion mode identification.The core letter of SVMs is used as by the use of RBF
Number.The global optimum that each subclassification supporting vector machine model is obtained by genetic algorithm (Genetic Algorithm, GA) punishes
Penalty parameter C and nuclear parameter γ.Shown in improved binary-tree support vector machine algorithm flow Fig. 5.
A part of characteristic vector is randomly selected to be used to train ISVM-BT to obtain optimal BT hierarchies, and with remaining
Characteristic vector verifies ISVM-BT classification performance.With the data instance of sample 1,20 samples are randomly selected from every group data set
This obtains optimal hierarchy and is illustrated in fig. 6 shown below as training set, remaining 20 samples as test set
Then globally optimal solution, setting C (0.1,1000) and γ are sought to SVM punishment parameters C and kernel function γ according to GA
(0.001,10) and globally optimal solution is obtained by 10 times of cross validation methods, classification results are as shown in table 1.
The classification results of table 1
Higher discrimination is realized for this seven actions of these three sample sets as can be seen from Table 1, wherein on
Building and the downstairs discrimination of the two actions have reached 100%, and average recognition rate is up to 98.62%, and the present invention is in lower limb for this explanation
There is higher discrimination and reliability in recognizing model of movement.
In order to verify fusion myoelectricity and acceleration identifying schemes and the simple quality for relying on myoelectricity, according to above-mentioned data processing
Method, surface electromyogram signal are as shown in table 2 to the classifying quality of seven actions.As can be seen from the table, it is single to rely on surface flesh
Electric signal to upstairs, downstairs and walk these three action discrimination it is higher, and to squat station, crouchings of stand, sit station and stand seat this four compare
Compared with similar movement discrimination than relatively low, compare Tables 1 and 2 and can be seen that the discriminations for introducing this seven actions after acceleration are obvious
Improve, wherein to crouching station, crouching of standing, sitting to stand and stand and sitting the most obvious of this four similar movement discriminations raisings.
The classification results of 2 single surface electromyogram signal of table
Claims (2)
1. merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal, it is characterised in that:This method is included such as
Lower step:
Step 1. obtains human body lower limbs surface electromyogram signal, acceleration signal;
The surface electromyogram signal that step 1 obtains is decomposed into multiple multiplicative functions by step 2. with local mean value decomposition algorithm;
Step 3. according to the average Euclidean distances for characterizing different action separation properties, determine local mean value decomposition algorithm decompose after the
The multiple dimensioned arrangement entropy of one multiplicative function, the multiple dimensioned arrangement entropy for extracting first multiplicative function are special as surface electromyogram signal
Sign;
Step 4. application Laplce's weight feature selecting algorithm calculates the weight of different scale entropy, by Laplce's weighted value most
Four high Scale Entropies form four dimensional feature vectors;
Four dimensional feature vectors and the ordering entropy composition 7 degree of freedom characteristic vector of 3-axis acceleration that step 5. is obtained step 4;
Step 6. inputs 7 degree of freedom characteristic vector according to average sample distribution between Euclidean distance and class in class and improved binary tree
SVMs carries out lower limb motion mode identification;The kernel function of SVMs is used as by the use of RBF;Calculated by heredity
Method obtains global optimum the punishment parameter C and nuclear parameter γ of each subclassification supporting vector machine model;
Described binary-tree support vector machine algorithmic procedure is as follows:
(1) average Euclidean distance AVI, weight coefficient k average between Euclidean distance AV and class in class are obtained by training datan=2n, -4
< < n < < 4, n is integer;
(2) for given knCalculate Separatory measure II, j, Separatory measure can be by average Euclidean distance AV in classi、AVjWith
Euclidean distance AVI between average classI, jI.e. I is obtained with weight coefficient kI, j=AVII, j+k(AVi+AVj), i, j=1,2 ..., N, i ≠
J, it is as follows that symmetrical matrix SI is constructed by Separatory measure
(3) what the hierarchy of binary tree was often gone according to matrix SI sorts with value;
(4) weights k is changednAnd repeat step 6- (2) and step 6- (3) generates a series of binary tree hierarchies;
(5) a binary tree hierarchy and weights k are selectedn, establish SVM sub-classifiers;Need to produce k-1 for k class problems
Sub-classifier, test data is classified, calculate the nicety of grading of grader;
(6) n=n+1 repeat steps 6- (5) is made untill n=4;
(7) according to all weights knCorresponding nicety of grading determines optimal binary tree hierarchy.
2. the lower limb motion mode recognition methods of fusion surface myoelectric according to claim 1 and acceleration signal, it is special
Sign is:Step 1 detailed process is as follows:
Four pieces of selected tibialis anterior, gastrocnemius, rectus femoris, semitendinosus muscle are obtained as signal source by myoelectricity acquisition system
Electromyographic signal;One 3-axis acceleration sensor node is fixed on tester's waist, captures and records the original of the axle of x, y, z three
Acceleration movement data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711013730.5A CN107832686A (en) | 2017-10-26 | 2017-10-26 | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711013730.5A CN107832686A (en) | 2017-10-26 | 2017-10-26 | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107832686A true CN107832686A (en) | 2018-03-23 |
Family
ID=61649396
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711013730.5A Pending CN107832686A (en) | 2017-10-26 | 2017-10-26 | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107832686A (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108542381A (en) * | 2018-04-11 | 2018-09-18 | 京东方科技集团股份有限公司 | A kind of data processing method and device |
CN108958474A (en) * | 2018-05-29 | 2018-12-07 | 西北工业大学 | A kind of action recognition multi-sensor data fusion method based on Error weight |
CN109192007A (en) * | 2018-09-21 | 2019-01-11 | 杭州电子科技大学 | A kind of AR sign Language Recognition Method and teaching method based on myoelectricity motion perception |
CN109558911A (en) * | 2018-12-26 | 2019-04-02 | 杭州电子科技大学 | Electromyography signal Feature fusion based on genetic algorithm broad sense canonical correlation analysis |
CN109635638A (en) * | 2018-10-31 | 2019-04-16 | 中国科学院计算技术研究所 | For the feature extracting method and system of human motion, recognition methods and system |
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 |
CN110025308A (en) * | 2019-04-09 | 2019-07-19 | 澳门大学 | A kind of cardioelectric characteristic extracting process, the heart clap recognition methods and device |
CN110334734A (en) * | 2019-05-31 | 2019-10-15 | 宁波中车时代传感技术有限公司 | A kind of intelligent sensing fusion method based on meta-learn technology |
CN110464517A (en) * | 2019-08-16 | 2019-11-19 | 杭州电子科技大学 | Electromyography signal recognition methods based on Wavelet-Weighted arrangement entropy |
CN111310812A (en) * | 2020-02-06 | 2020-06-19 | 佛山科学技术学院 | Data-driven hierarchical human activity recognition method and system |
CN111803249A (en) * | 2020-07-24 | 2020-10-23 | 北京海益同展信息科技有限公司 | Method, device and equipment for selecting signal channel of prosthesis and storage medium |
CN112461546A (en) * | 2020-10-27 | 2021-03-09 | 江苏大学 | Construction method and diagnosis method of pump bearing fault diagnosis model based on improved binary tree support vector machine |
CN113456065A (en) * | 2021-08-10 | 2021-10-01 | 长春理工大学 | Limb action recognition method, device and system and readable storage medium |
CN117281667A (en) * | 2023-11-09 | 2023-12-26 | 浙江强脑科技有限公司 | Motion pattern recognition method and device, intelligent artificial limb, terminal and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101980251A (en) * | 2010-11-23 | 2011-02-23 | 中国矿业大学 | Remote sensing classification method for binary tree multi-category support vector machines |
CN102722643A (en) * | 2012-05-28 | 2012-10-10 | 天津大学 | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb |
CN105678261A (en) * | 2016-01-07 | 2016-06-15 | 西安电子科技大学 | Supervised figure-based transductive data dimension-descending method |
CN107101813A (en) * | 2017-04-26 | 2017-08-29 | 河北工业大学 | A kind of frame-type circuit breaker mechanical breakdown degree assessment method based on vibration signal |
-
2017
- 2017-10-26 CN CN201711013730.5A patent/CN107832686A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101980251A (en) * | 2010-11-23 | 2011-02-23 | 中国矿业大学 | Remote sensing classification method for binary tree multi-category support vector machines |
CN102722643A (en) * | 2012-05-28 | 2012-10-10 | 天津大学 | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb |
CN105678261A (en) * | 2016-01-07 | 2016-06-15 | 西安电子科技大学 | Supervised figure-based transductive data dimension-descending method |
CN107101813A (en) * | 2017-04-26 | 2017-08-29 | 河北工业大学 | A kind of frame-type circuit breaker mechanical breakdown degree assessment method based on vibration signal |
Non-Patent Citations (2)
Title |
---|
刘磊: "基于多源信息的步态识别算法研究", 《中国优秀博士学位论文全文数据库信息科技辑》 * |
赵海洋: "改进二叉树支持向量机及其故障诊断方法研究", 《振动工程学报》 * |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108542381A (en) * | 2018-04-11 | 2018-09-18 | 京东方科技集团股份有限公司 | A kind of data processing method and device |
CN108958474A (en) * | 2018-05-29 | 2018-12-07 | 西北工业大学 | A kind of action recognition multi-sensor data fusion method based on Error weight |
CN109192007A (en) * | 2018-09-21 | 2019-01-11 | 杭州电子科技大学 | A kind of AR sign Language Recognition Method and teaching method based on myoelectricity motion perception |
CN109635638A (en) * | 2018-10-31 | 2019-04-16 | 中国科学院计算技术研究所 | For the feature extracting method and system of human motion, recognition methods and system |
CN109635638B (en) * | 2018-10-31 | 2021-03-09 | 中国科学院计算技术研究所 | Feature extraction method and system and recognition method and system for human body motion |
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 |
CN109558911A (en) * | 2018-12-26 | 2019-04-02 | 杭州电子科技大学 | Electromyography signal Feature fusion based on genetic algorithm broad sense canonical correlation analysis |
CN110025308A (en) * | 2019-04-09 | 2019-07-19 | 澳门大学 | A kind of cardioelectric characteristic extracting process, the heart clap recognition methods and device |
CN110025308B (en) * | 2019-04-09 | 2021-09-10 | 澳门大学 | Electrocardio feature extraction method, heart beat identification method and device |
CN110334734A (en) * | 2019-05-31 | 2019-10-15 | 宁波中车时代传感技术有限公司 | A kind of intelligent sensing fusion method based on meta-learn technology |
CN110464517A (en) * | 2019-08-16 | 2019-11-19 | 杭州电子科技大学 | Electromyography signal recognition methods based on Wavelet-Weighted arrangement entropy |
CN110464517B (en) * | 2019-08-16 | 2021-09-07 | 杭州电子科技大学 | Electromyographic signal identification method based on wavelet weighted arrangement entropy |
CN111310812A (en) * | 2020-02-06 | 2020-06-19 | 佛山科学技术学院 | Data-driven hierarchical human activity recognition method and system |
CN111310812B (en) * | 2020-02-06 | 2023-04-28 | 佛山科学技术学院 | Hierarchical human body activity recognition method and system based on data driving |
CN111803249A (en) * | 2020-07-24 | 2020-10-23 | 北京海益同展信息科技有限公司 | Method, device and equipment for selecting signal channel of prosthesis and storage medium |
CN112461546A (en) * | 2020-10-27 | 2021-03-09 | 江苏大学 | Construction method and diagnosis method of pump bearing fault diagnosis model based on improved binary tree support vector machine |
CN113456065A (en) * | 2021-08-10 | 2021-10-01 | 长春理工大学 | Limb action recognition method, device and system and readable storage medium |
CN113456065B (en) * | 2021-08-10 | 2022-08-26 | 长春理工大学 | Limb action recognition method, device and system and readable storage medium |
CN117281667A (en) * | 2023-11-09 | 2023-12-26 | 浙江强脑科技有限公司 | Motion pattern recognition method and device, intelligent artificial limb, terminal and storage medium |
CN117281667B (en) * | 2023-11-09 | 2024-04-09 | 浙江强脑科技有限公司 | Motion pattern recognition method and device, intelligent artificial limb, terminal and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107832686A (en) | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal | |
Tang et al. | Impact of load variation on joint angle estimation from surface EMG signals | |
Rasool et al. | Real-time task discrimination for myoelectric control employing task-specific muscle synergies | |
CN103054585B (en) | Biological motion information based upper limb shoulder elbow wrist joint motion function evaluation method | |
CN104107042A (en) | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine | |
CN103610466A (en) | EMG fall identification method based on EMD permutation entropy | |
Ahsan et al. | Hand motion detection from EMG signals by using ANN based classifier for human computer interaction | |
CN102722643A (en) | Multi-information merging movement mode identification method used for control of artificial lower surface electromyogram limb | |
CN108681685A (en) | A kind of body work intension recognizing method based on human body surface myoelectric signal | |
Wang et al. | A portable artificial robotic hand controlled by EMG signal using ANN classifier | |
CN110363152A (en) | A kind of artificial leg road conditions recognition methods based on surface electromyogram signal | |
Li et al. | Gait recognition based on EMG with different individuals and sample sizes | |
CN109394472A (en) | A kind of healing robot motion intention recognition methods based on neural network classifier | |
Yu et al. | Attenuating the impact of limb position on surface EMG pattern recognition using a mixed-LDA classifier | |
Attenberger et al. | Modeling and visualization of classification-based control schemes for upper limb prostheses | |
Kadrolkar et al. | Intent recognition of torso motion using wavelet transform feature extraction and linear discriminant analysis ensemble classification | |
CN110032987B (en) | Surface electromyographic signal classification method based on cerebellar neural network model | |
Nougarou et al. | Muscle activity distribution features extracted from HD sEMG to perform forearm pattern recognition | |
CN109634419A (en) | The recognition methods of healing robot motion intention and its computer readable storage medium | |
Veer | A flexible approach for segregating physiological signals | |
Zhang et al. | Activity recognition of the torso based on surface electromyography for exoskeleton control | |
Xia et al. | Predicting Free Achilles Tendon Strain From Motion Capture Data Using Artificial Intelligence | |
Tomczyński et al. | Hand gesture-based interface with multichannel sEMG band enabling unknown gesture discrimination | |
Sun et al. | A fault-tolerant algorithm to enhance generalization of EMG-based pattern recognition for lower limb movement | |
CN102521508A (en) | Adaptive neural fuzzy muscle modeling method under functional electrical stimulation |
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: 20180323 |