CN112244851A - Muscle movement recognition method and surface electromyogram signal acquisition device - Google Patents
Muscle movement recognition method and surface electromyogram signal acquisition device Download PDFInfo
- Publication number
- CN112244851A CN112244851A CN202011268602.7A CN202011268602A CN112244851A CN 112244851 A CN112244851 A CN 112244851A CN 202011268602 A CN202011268602 A CN 202011268602A CN 112244851 A CN112244851 A CN 112244851A
- Authority
- CN
- China
- Prior art keywords
- signal
- filtered
- electromyographic
- muscle movement
- signals
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000033001 locomotion Effects 0.000 title claims abstract description 47
- 210000003205 muscle Anatomy 0.000 title claims abstract description 40
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims abstract description 11
- 230000005540 biological transmission Effects 0.000 claims abstract description 10
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 7
- 239000002245 particle Substances 0.000 claims description 71
- 239000010410 layer Substances 0.000 claims description 28
- 238000001914 filtration Methods 0.000 claims description 26
- 239000011159 matrix material Substances 0.000 claims description 24
- 239000002344 surface layer Substances 0.000 claims description 18
- 239000004744 fabric Substances 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 10
- 230000002401 inhibitory effect Effects 0.000 claims description 8
- 239000000758 substrate Substances 0.000 claims description 7
- 230000008030 elimination Effects 0.000 claims description 6
- 238000003379 elimination reaction Methods 0.000 claims description 6
- 239000000463 material Substances 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- BQCADISMDOOEFD-UHFFFAOYSA-N Silver Chemical compound [Ag] BQCADISMDOOEFD-UHFFFAOYSA-N 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 229910052709 silver Inorganic materials 0.000 claims description 4
- 239000004332 silver Substances 0.000 claims description 4
- 239000004677 Nylon Substances 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 238000002567 electromyography Methods 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 229920001778 nylon Polymers 0.000 claims description 3
- 230000006978 adaptation Effects 0.000 claims description 2
- 230000003139 buffering effect Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 10
- 238000007637 random forest analysis Methods 0.000 description 8
- 230000009471 action Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000003066 decision tree Methods 0.000 description 3
- 238000000354 decomposition reaction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 210000001087 myotubule Anatomy 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 230000036982 action potential Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- 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
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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
- A61B5/7253—Details of waveform analysis characterised by using transforms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2562/00—Details of sensors; Constructional details of sensor housings or probes; Accessories for sensors
- A61B2562/02—Details of sensors specially adapted for in-vivo measurements
- A61B2562/0209—Special features of electrodes classified in A61B5/24, A61B5/25, A61B5/283, A61B5/291, A61B5/296, A61B5/053
- A61B2562/0215—Silver or silver chloride containing
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Biophysics (AREA)
- Heart & Thoracic Surgery (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Power Engineering (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to a muscle movement identification method and a surface electromyogram signal acquisition device, wherein the method comprises the following steps: collecting historical surface electromyographic signals by using a surface electromyographic signal collecting device; processing the historical surface electromyographic signals to obtain a surface electromyographic signal characteristic subset; constructing a convolutional neural network model; training the convolutional neural network model through the surface electromyographic signal feature subset to obtain a muscle motion recognition model; and identifying the muscle movement according to the muscle movement identification model. The device comprises: the system comprises an electromyographic signal sensor, a low-pass filter, an operational amplifier, an analog-to-digital converter and a wireless transmission module. According to the invention, the surface electromyogram signal characteristic subset is obtained by processing the historical surface electromyogram signal, and the muscle movement is identified by using the muscle movement identification model obtained by training the surface electromyogram signal characteristic subset, so that the identification process is quicker and more accurate.
Description
Technical Field
The invention relates to the technical field of muscle movement detection, in particular to a muscle movement identification method and a surface electromyographic signal acquisition device.
Background
Surface electromyography (sEMG) is the superposition of action potentials of motor units in muscle fibers on time and space, the nervous system controls the activity (contraction or relaxation) of muscles, and different muscle fiber motor units on the surface skin generate different signals at the same time, and the signals contain a series of abundant pathological/physiological information. At present, sEMG is widely applied to the fields of clinical medicine, human-computer efficiency, rehabilitation medicine, sports science, prosthetic limb application, gesture action recognition and the like. However, the rigid electrodes such as silver have the problems that the contact impedance of the skin electrode is large, the skin electrode is easily interfered by power frequency signals, the contact area of the skin surface is uneven and the like in the process of acquiring the surface electromyographic signals; in addition, the problems of strong external interference of the acquired sEMG signal, complex multi-channel signal processing, difficult feature vector selection and the like limit further development and large-scale application of the sEMG signal.
Disclosure of Invention
The invention aims to provide a muscle movement identification method and a surface electromyogram signal acquisition device so as to quickly and accurately acquire surface electromyogram signals and identify muscle movement.
In order to achieve the purpose, the invention provides the following scheme:
a surface electromyogram signal acquisition device, comprising:
the electromyographic signal sensor is used for acquiring an initial surface electromyographic signal;
the low-pass filter is connected with the electromyographic signal sensor and used for filtering the initial surface electromyographic signal to obtain a filtered signal;
the operational amplifier is connected with the low-pass filter and used for amplifying the filtered signal to obtain an amplified signal;
the analog-to-digital converter is connected with the operational amplifier and is used for converting the amplified signal into a digital signal to obtain a surface electromyographic signal;
and the wireless transmission module is connected with the analog-to-digital converter and is used for transmitting the surface electromyographic signals.
Optionally, the electromyographic signal sensor comprises a touch surface layer, a motion trail inhibiting layer, a fabric conductive layer and a substrate which are arranged in sequence from top to bottom; the touch surface layer is used for contacting with the surface of skin to be collected and collecting an initial surface electromyographic signal; the motion trail inhibiting layer is used for buffering the interaction between the contact surface layer and the surface of the skin to be collected; the fabric conducting layer is used for transmitting the initial surface electromyographic signals.
Optionally, the contact surface layer is a knitted silver conductive fabric, the motion trail inhibiting layer is a conductive sponge, and the substrate is made of an elastic nylon material.
Optionally, the electromyographic signal sensor further comprises:
the insulating shielding layers are arranged on two sides of the contact surface layer and used for isolating the surface of the skin to be collected from the fabric conducting layer; the insulating shielding layer is made of insulating fabric materials with hollow centers.
Optionally, the low-pass filter is an RC low-pass filter; the operational amplifier and the analog-to-digital converter are ADS1299 chips; the wireless transmission module is a WIFI module integrated in the CC3200 chip.
A muscle movement recognition method applying the surface electromyogram signal acquisition device comprises the following steps:
acquiring historical surface electromyographic signals;
processing the historical surface electromyographic signals to obtain a surface electromyographic signal characteristic subset;
constructing a convolutional neural network model;
training the convolutional neural network model through the surface electromyographic signal feature subset to obtain a muscle motion recognition model;
and identifying the muscle movement according to the muscle movement identification model.
Optionally, the processing the historical surface electromyogram signal to obtain a surface electromyogram signal feature subset includes:
carrying out normalization processing on the historical surface electromyographic signals to obtain normalized signals;
filtering the normalized signal to obtain a filtered signal;
extracting time domain characteristics of the filtered signals and constructing a characteristic matrix;
filtering the feature matrix to obtain a filtered feature matrix;
and obtaining a surface electromyogram signal characteristic subset by utilizing a random forest-recursive characteristic elimination algorithm according to the filtered characteristic matrix.
Optionally, the filtering processing is performed on the normalized signal to obtain a filtered signal, specifically:
carrying out discrete wavelet change on the normalized signal to obtain a high-frequency signal coefficient and a low-frequency signal coefficient;
filtering the high-frequency signal coefficient according to a preset threshold value to obtain a filtered high-frequency signal coefficient;
and performing signal reconstruction by using the low-frequency signal coefficient and the filtered high-frequency signal coefficient to obtain a filtered signal.
Optionally, the filtering processing is performed on the feature matrix, specifically using a formula:
filtering the feature matrix; wherein, yiIs the ith value in the filtered feature matrix, N is the filter order, fjFor the jth value extracted from the feature matrix, fjThe sum of the sums of (1),
q is a proportionality coefficient.
Optionally, the filter order and the scaling factor are calculated using a particle swarm algorithm; wherein, the particle swarm algorithm comprises the following calculation processes:
initializing a particle number using the filter order and the scaling factor;
calculating a particle adaptation value according to the position of the particle;
updating the optimal positions of the particles, the optimal positions of the particle swarms and the speed and the positions of the particles according to the adaptive values of the particles, the optimal positions of the particles and the optimal positions of the particle swarms under the number of the particles;
outputting final particles according to the updated optimal positions of the particle swarms; the final particles are the filter order and the scaling factor.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention discloses a muscle movement identification method and a surface electromyogram signal acquisition device, wherein the method comprises the following steps: collecting historical surface electromyographic signals by using a surface electromyographic signal collecting device; processing the historical surface electromyographic signals to obtain a surface electromyographic signal characteristic subset; constructing a convolutional neural network model; training the convolutional neural network model through the surface electromyographic signal feature subset to obtain a muscle motion recognition model; and identifying the muscle movement according to the muscle movement identification model. The device comprises: the electromyographic signal sensor is used for acquiring an initial surface electromyographic signal; the low-pass filter is connected with the electromyographic signal sensor and used for filtering the initial surface electromyographic signal to obtain a filtered signal; the operational amplifier is connected with the low-pass filter and used for amplifying the filtered signal to obtain an amplified signal; the analog-to-digital converter is connected with the operational amplifier and is used for converting the amplified signal into a digital signal to obtain a surface electromyographic signal; and the wireless transmission module is connected with the analog-to-digital converter and is used for transmitting the surface electromyographic signals. According to the invention, the surface electromyogram signal characteristic subset is obtained by processing the historical surface electromyogram signal, and the muscle movement is identified by using the muscle movement identification model obtained by training the surface electromyogram signal characteristic subset, so that the identification process is quicker and more accurate.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a structural diagram of a surface electromyogram signal acquisition device provided in embodiment 1 of the present invention;
fig. 2 is a circuit diagram of an RC low pass circuit provided in embodiment 1 of the present invention;
fig. 3 is a structural diagram of an electromyographic signal sensor provided in embodiment 1 of the present invention;
fig. 4 is a flowchart of a muscle movement identification method according to embodiment 2 of the present invention;
fig. 5 is a flowchart of obtaining a surface electromyogram signal feature subset according to embodiment 2 of the present invention;
fig. 6 is a flowchart of filtering a normalized signal according to embodiment 2 of the present invention;
FIG. 7 is a signal diagram before filtering according to embodiment 2 of the present invention;
FIG. 8 is a diagram of filtered signals provided in embodiment 2 of the present invention;
fig. 9 is a schematic diagram of a particle swarm algorithm process provided in embodiment 2 of the present invention;
fig. 10 is a flowchart of a random forest-recursive feature elimination algorithm provided in embodiment 2 of the present invention.
Description of the symbols: 1-contact surface layer, 2-motion track inhibiting layer, 3-substrate, 4-insulating shielding layer and 5-fabric conducting layer.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a muscle movement identification method and a surface electromyogram signal acquisition device so as to quickly and accurately acquire surface electromyogram signals and identify muscle movement.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Example 1
Fig. 1 is a structural diagram of a surface electromyogram signal acquisition device provided in embodiment 1 of the present invention, and as shown in fig. 1, the surface electromyogram signal acquisition device includes:
and the electromyographic signal sensor is used for acquiring an initial surface electromyographic signal.
And the low-pass filter is connected with the electromyographic signal sensor and is used for filtering the initial surface electromyographic signal to obtain a filtered signal.
And the operational amplifier is connected with the low-pass filter and used for amplifying the filtered signal to obtain an amplified signal.
And the analog-to-digital converter is connected with the operational amplifier and is used for converting the amplified signal into a digital signal to obtain a surface electromyographic signal.
And the wireless transmission module is connected with the analog-to-digital converter and is used for transmitting the surface muscle electrical signal.
In this embodiment, the surface electromyogram signal acquisition device further includes a controller, connected to the electromyogram signal sensor, the low-pass filter, the operational amplifier, the analog-to-digital converter, and the wireless transmission module, for controlling operations of the electromyogram signal sensor, the low-pass filter, the operational amplifier, the analog-to-digital converter, and the wireless transmission module. Wherein the control module is a CC3200 chip.
In this embodiment, the operational amplifier and the analog-to-digital converter are ADS1299 chips. The wireless transmission module is a WIFI module integrated inside the CC3200 chip. The low-pass filter is an RC low-pass filter, and fig. 2 is a circuit diagram of an RC low-pass circuit provided in embodiment 1 of the present invention.
Fig. 3 is a structural diagram of an electromyographic signal sensor according to embodiment 1 of the present invention, and as shown in fig. 3, the electromyographic signal sensor includes a touch surface layer 1, a motion trajectory inhibiting layer 2, a fabric conductive layer 5, and a substrate 3, which are sequentially disposed from top to bottom. The touch surface layer 1 is used for contacting with the surface of skin to be collected and collecting the initial surface electromyographic signals. The motion trajectory suppression layer 2 serves to cushion the interaction between the touch surface layer 1 and the skin surface to be harvested. The fabric conductive layer 5 is used for transmitting the initial surface electromyographic signals and can also be used for shielding high-frequency interference signals. Specifically, the electromyographic signal sensor further comprises insulating shielding layers 4 arranged on two sides of the touch surface layer 1 and used for isolating the surface of the skin to be collected from the fabric conducting layer 5.
The contact surface layer 1 is made of knitted silver conductive cloth, so that the problem that a rigid electrode contact area is uneven is solved, and the contact surface layer has good conductivity and extremely high sensitivity. The motion track restraining layer 2 is a conductive sponge. The substrate 3 is an elastic nylon material. The insulating shielding layer 4 is made of insulating fabric material with a hollow center, and has a certain size difference with the contact surface layer 1 to form a supporting plane, so that the skin surface to be collected is prevented from contacting the fabric conductive layer 5 to generate interference signals.
Example 2
Fig. 4 is a flowchart of a muscle movement identification method according to embodiment 2 of the present invention, and as shown in fig. 4, the muscle movement identification method includes:
step 101: acquiring historical surface electromyographic signals. Wherein the historical surface electromyogram signal is acquired based on the surface electromyogram signal acquisition apparatus in embodiment 1.
Step 102: and processing the historical surface electromyographic signals to obtain a surface electromyographic signal characteristic subset.
Step 103: and constructing a convolutional neural network model.
Step 104: and training the convolutional neural network model through the surface electromyographic signal feature subset to obtain a muscle motion recognition model.
Step 105: and identifying the muscle movement according to the muscle movement identification model.
Fig. 5 is a flowchart of obtaining a surface electromyogram signal feature subset according to embodiment 2 of the present invention, and as shown in fig. 5, step 102 specifically includes:
step 1021: carrying out normalization processing on the historical surface electromyographic signals to obtain normalized signals; specifically, the formula is utilized:
carrying out normalization processing on the historical surface electromyographic signals; wherein, Xi' is normalized surface electromyographic signal, XiIn order to be a historical surface electromyographic signal,
is XiThe average value of (a) of (b),
1,2, n, i is the ith value, n is the total number, σ2Is the variance of the received signal and the received signal,
step 1022: filtering the normalized signal to obtain a filtered signal; the method specifically comprises the following steps: and carrying out discrete wavelet change on the normalized signal to obtain a high-frequency signal coefficient and a low-frequency signal coefficient. And filtering the high-frequency signal coefficient according to a preset threshold value to obtain a filtered high-frequency signal coefficient. And performing signal reconstruction by using the low-frequency signal coefficient and the filtered high-frequency signal coefficient to obtain a filtered signal.
FIG. 6 is a flow chart of filtering the normalized signal according to embodiment 2 of the present invention, as shown in FIG. 5, in the space Vj＝Vj-1+Wj-1Upper representation of normalized posterior surface electromyographic signal Xi', as shown in formula (2), for each at VjThe signal in space can be represented by two basis functions.
Wherein A is1(k) And D1(k) Are two coefficients of a scale metric space j-1, coefficient A from the j space0C is a constant, phij,k(t) is Xi' decomposition amount in space j, phij-1,k(t) and ωj-1,k(t) is the amount of decomposition in space j-1. A. the0(k) Is the entire spatial coefficient.
A0(k) Is decomposed into coefficients A1(k) And D1(k) The process of (2) is as follows:
wherein A is1(k) Corresponding precision factor, D1(k) Corresponding to the coarse coefficient, h0Low pass filter coefficient, h1High-pass filter coefficients, n denotes the filter order, and k denotes the kth data.
For coefficient D after decomposition1(k) The filtering threshold ξ is set as shown in equation (4):
where N is the total number of signals, m is media (| D)1(k) |)/0.6745, m is a coefficient, and Median () is a Median function.
To D1(k) Filtering when D1(k) When the value in (D) is less than the threshold value xi, setting to zero, and when D is less than the threshold value xi, setting to zero1(k) When the value of (1) is greater than or equal to the threshold value xi, the value is kept unchanged, namely:
and reconstructing the filtered signal according to the formula (2) to obtain the filtered signal. A comparison of the signals before and after filtering is shown in fig. 7-8.
Step 1023: extracting time domain characteristics of the filtered signals and constructing a characteristic matrix; the time domain features include root mean square, variance, wavelength length, number of zero-crossing points, average absolute value, maximum fractal length, average energy and autoregressive coefficient.
The root mean square RMS calculation is as follows:
wherein N is the total number of signals, YiIs the ith signal in the filtered signal Y.
The variance VAR calculation is as follows:
the wavelength length WL is calculated as follows:
wherein, Yi+1Is the ith signal in the filtered signal Y.
The zero crossing number ZC calculation formula is as follows:
where τ is a threshold for avoiding low-level noise and is set to 0.5, sgn (x) is a sign function for determining whether a single zero-crossing point occurs.
The average absolute value MAV is calculated as follows:
the maximum fractal length MLF calculation formula is as follows:
the average energy AP is calculated as follows:
the autoregressive coefficient calculation formula is as follows:
wherein alpha is an autoregressive coefficient, p is the order of the autoregressive coefficient, e is a residual error, YkIs a time series with the number of sequences k, k being the number of sequences.
Step 1024: and filtering the feature matrix to obtain a filtered feature matrix.
When some actions of the electromyographic signals are collected, it is possible that interference such as inconsistency of execution action strength inevitably causes different amplitudes of the electromyographic signals of the same action, so that isolated points or abnormal values exist in different windows. In other words, considering the existence of isolated points/abnormal values, each point is replaced by multiplying the previous and next N values by a proportional coefficient decreasing with time, specifically by using the formula:
filtering the feature matrix; wherein, yiIs the ith value in the filtered feature matrix, N is the filter order, fjFor the jth value extracted from the feature matrix, fjThe sum of the sums of (1),
q is a proportionality coefficient. Filter order sum ratio using particle swarm optimizationExample coefficients are optimized, and the specific process is as follows:
the number of particles is initialized with the filter order and the scaling factor.
A particle fitness value is calculated based on the position of the particle.
And updating the optimal positions of the particles, the optimal positions of the particle swarm and the speed and the positions of the particles according to the adaptive values of the particles, the optimal positions of the particles and the optimal positions of the particle swarm under the number of the particles.
Outputting final particles according to the updated optimal positions of the particle swarms; the final particles are the filter order and the scaling factor.
Fig. 9 is a schematic diagram of a particle swarm algorithm process provided in embodiment 2 of the present invention, and as shown in fig. 9, a specific optimization process is as follows:
initializing random particles, setting the maximum iteration number 800, setting the maximum speed Vmax of the particles to be 5, setting the position information to be the whole search space, randomly initializing the speed and the position in a speed interval and the search space, setting the particle swarm size to be M (namely different combinations of filter orders and proportionality coefficients), and randomly initializing one flying speed for each particle.
The update speed and position formula of the particle is as follows:
wherein i is the ith particle, i is 1,21，N1Is the total number of particles, viFor particle velocity, rand () for generating random number, xiIs the position of the particle, c1And c2Is a learning factor and is set to 2.
The particle update rate comprises three components:
(1) the first fraction of particles is the previous velocity.
(2) The second part is the cognitive part, which represents the thought of the particle itself, i.e. the distance of the current position of the particle from the optimal position.
(3) The third part is a social part and represents information sharing and cooperation among the particles, namely, the optimal positions among the particles and the groups.
And (3) calculating a particle adaptive value according to the following calculation formula:
lny＝(cos(2Πx[0])+cos(2Πx[1]))/2-2.71289 (16)
where y is the fitness value and x is the particle position.
A population of particles (including random positions and velocities) is initialized.
The fitness value of each particle was evaluated and evaluated.
For each particle, its fitness value is compared to its past best position pbest, and if better, it is taken as the current best position pbest. (pbest is the historical best position)
For each particle, its fitness value is compared to the best position it passes through, gbest, and if better, it is taken as the current best position gbest. (gbest is the global optimum position)
The particle velocity and position are adjusted according to equation (15).
It is determined whether an end condition is reached. The end condition is that the maximum number of iterations 500 is reached or that the global optimum position meets the minimum limit J.
If not, re-evaluating and evaluating the adaptive value of each particle; if yes, outputting final particles, wherein the final particles are the filter order and the proportionality coefficient.
Step 1025: and obtaining a surface electromyogram signal characteristic subset by utilizing a random forest-recursive characteristic elimination algorithm according to the filtered characteristic matrix. The method specifically comprises the following steps:
extracting a plurality of samples from original samples by using a bootstrap resampling method, constructing a decision tree for each bootstrap sample, forming a random forest by all the decision trees, calculating feature importance in a regression model, introducing backward iterative feature evaluation, calculating feature importance of the remaining features by using a random forest algorithm again after deleting the features with small feature importance until only one feature is left at last, and selecting the most feature set according to a correlation coefficient and a root mean square error.
The detailed process is as follows:
(1) the number of samples of the filtered matrix S is n, b sample subsets are randomly extracted in a replacement mode by using bootstrap sampling, b regression trees are constructed according to the sample subsets, samples which are not extracted each time bootstrap sampling form b extra-bag data, and the extra-bag data form test samples of random forests.
(2) Setting the characteristic number of an original sample set as p, randomly extracting m variables (m < ═ p) at each node of each regression tree as alternative variables, then selecting an optimal branch according to a certain criterion, and determining the maximum growth of each decision tree under different conditions;
(3) integrating the b regression trees generated in the step (1) into a random forest regression model, and evaluating the effect of the random forest regression model by using residual Mean Square Error (MSE) predicted by using data outside a bag, wherein the calculation formula of the MSE is as follows:
wherein s isiIs the actual value of the dependent variable in the out-of-bag data,
and predicting the data outside the bag for the random forest.
(4) And calculating an average descending MSE value through the residual mean square of the data prediction outside the bag, wherein the importance of the variable in the random forest regression can be measured by the average descending MSE value, and the larger the value is, the more important the characteristic is.
(5) And after the average descending MSE value is obtained through calculation, firstly deleting the feature with the minimum feature importance degree according to the principle of backward iteration, then repeating the step (1) to the step (4) on the remaining features, gradually deleting the features with the small importance degree until the last feature remains, and after the result is output, selecting the feature with the minimum root mean square error and the maximum correlation coefficient as the result of feature selection for forest biomass remote sensing estimation. Fig. 10 is a flowchart of a random forest-recursive feature elimination algorithm provided in embodiment 2 of the present invention.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
(1) the invention carries out twice filtering processing on the signal, and double denoising is carried out, thus greatly reducing the noise and improving the signal-to-noise ratio.
(2) The method adopts the RF-RFE (random forest-recursive feature elimination algorithm), the RF-RFE algorithm can reevaluate the current residual feature set in the iterative process when the feature selection is carried out, the score of each feature is adjusted in the iterative process, and the defect that the feature selection result of a single random forest needs to be tested repeatedly is overcome.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (10)
1. A surface electromyogram signal acquisition device, comprising:
the electromyographic signal sensor is used for acquiring an initial surface electromyographic signal;
the low-pass filter is connected with the electromyographic signal sensor and used for filtering the initial surface electromyographic signal to obtain a filtered signal;
the operational amplifier is connected with the low-pass filter and used for amplifying the filtered signal to obtain an amplified signal;
the analog-to-digital converter is connected with the operational amplifier and is used for converting the amplified signal into a digital signal to obtain a surface electromyographic signal;
and the wireless transmission module is connected with the analog-to-digital converter and is used for transmitting the surface electromyographic signals.
2. The surface electromyogram signal acquisition device of claim 1, wherein the electromyogram signal sensor comprises a touch surface layer, a motion trail inhibiting layer, a fabric conductive layer and a substrate which are arranged in sequence from top to bottom; the touch surface layer is used for contacting with the surface of skin to be collected and collecting an initial surface electromyographic signal; the motion trail inhibiting layer is used for buffering the interaction between the contact surface layer and the surface of the skin to be collected; the fabric conducting layer is used for transmitting the initial surface electromyographic signals.
3. The surface electromyogram signal acquisition device of claim 2, wherein the touch surface layer is a knitted silver conductive cloth, the motion trajectory inhibiting layer is a conductive sponge, and the substrate is an elastic nylon material.
4. The surface electromyographic signal acquisition device of claim 2, wherein the electromyographic signal sensor further comprises:
the insulating shielding layers are arranged on two sides of the contact surface layer and used for isolating the surface of the skin to be collected from the fabric conducting layer; the insulating shielding layer is made of insulating fabric materials with hollow centers.
5. The surface electromyogram signal acquisition device of claim 1, wherein the low-pass filter is an RC low-pass filter; the operational amplifier and the analog-to-digital converter are ADS1299 chips; the wireless transmission module is a WIFI module integrated in the CC3200 chip.
6. A muscle movement recognition method, wherein the surface electromyogram signal acquisition apparatus of any one of claims 1 to 5 is applied, the method comprising:
acquiring historical surface electromyographic signals;
processing the historical surface electromyographic signals to obtain a surface electromyographic signal characteristic subset;
constructing a convolutional neural network model;
training the convolutional neural network model through the surface electromyographic signal feature subset to obtain a muscle motion recognition model;
and identifying the muscle movement according to the muscle movement identification model.
7. The muscle movement recognition method according to claim 6, wherein the processing of the historical surface electromyography signals to obtain a surface electromyography signal feature subset comprises:
carrying out normalization processing on the historical surface electromyographic signals to obtain normalized signals;
filtering the normalized signal to obtain a filtered signal;
extracting time domain characteristics of the filtered signals and constructing a characteristic matrix;
filtering the feature matrix to obtain a filtered feature matrix;
and obtaining a surface electromyogram signal characteristic subset by utilizing a random forest-recursive characteristic elimination algorithm according to the filtered characteristic matrix.
8. The muscle movement recognition method according to claim 7, wherein the filtering processing is performed on the normalized signal to obtain a filtered signal, specifically:
carrying out discrete wavelet change on the normalized signal to obtain a high-frequency signal coefficient and a low-frequency signal coefficient;
filtering the high-frequency signal coefficient according to a preset threshold value to obtain a filtered high-frequency signal coefficient;
and performing signal reconstruction by using the low-frequency signal coefficient and the filtered high-frequency signal coefficient to obtain a filtered signal.
9. The muscle movement recognition method according to claim 7, wherein the feature matrix is filtered, specifically using a formula:
filtering the feature matrix; wherein, yiIs the ith value in the filtered feature matrix, N is the filter order, fjFor the jth value extracted from the feature matrix, fjThe sum of the sums of (1),
q is a proportionality coefficient.
10. The muscle movement identification method according to claim 9, wherein the filter order and the scaling factor are calculated using a particle swarm algorithm; wherein, the particle swarm algorithm comprises the following calculation processes:
initializing a particle number using the filter order and the scaling factor;
calculating a particle adaptation value according to the position of the particle;
updating the optimal positions of the particles, the optimal positions of the particle swarms and the speed and the positions of the particles according to the adaptive values of the particles, the optimal positions of the particles and the optimal positions of the particle swarms under the number of the particles;
outputting final particles according to the updated optimal positions of the particle swarms; the final particles are the filter order and the scaling factor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011268602.7A CN112244851A (en) | 2020-11-13 | 2020-11-13 | Muscle movement recognition method and surface electromyogram signal acquisition device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011268602.7A CN112244851A (en) | 2020-11-13 | 2020-11-13 | Muscle movement recognition method and surface electromyogram signal acquisition device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112244851A true CN112244851A (en) | 2021-01-22 |
Family
ID=74265533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011268602.7A Pending CN112244851A (en) | 2020-11-13 | 2020-11-13 | Muscle movement recognition method and surface electromyogram signal acquisition device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112244851A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112932508A (en) * | 2021-01-29 | 2021-06-11 | 电子科技大学 | Finger activity recognition system based on arm electromyography network |
| CN113576483A (en) * | 2021-08-06 | 2021-11-02 | 中国科学院苏州生物医学工程技术研究所 | Wearable equipment |
| CN114931389A (en) * | 2022-04-27 | 2022-08-23 | 福州大学 | Electromyographic signal identification method based on residual error network and graph convolution network |
-
2020
- 2020-11-13 CN CN202011268602.7A patent/CN112244851A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112932508A (en) * | 2021-01-29 | 2021-06-11 | 电子科技大学 | Finger activity recognition system based on arm electromyography network |
| CN113576483A (en) * | 2021-08-06 | 2021-11-02 | 中国科学院苏州生物医学工程技术研究所 | Wearable equipment |
| CN114931389A (en) * | 2022-04-27 | 2022-08-23 | 福州大学 | Electromyographic signal identification method based on residual error network and graph convolution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112244851A (en) | Muscle movement recognition method and surface electromyogram signal acquisition device | |
| CN102622605B (en) | Surface electromyogram signal feature extraction and action pattern recognition method | |
| CN108304917A (en) | A kind of P300 signal detecting methods based on LSTM networks | |
| CN107832737A (en) | Electrocardiogram interference identification method based on artificial intelligence | |
| CN105769173A (en) | Electrocardiogram monitoring system with electrocardiosignal denoising function | |
| CN113111831A (en) | Gesture recognition technology based on multi-mode information fusion | |
| CN109009143B (en) | Method for predicting electrocardio information through human gait | |
| CN106127191B (en) | Brain electricity classification method based on WAVELET PACKET DECOMPOSITION and logistic regression | |
| Powar et al. | A novel pre-processing procedure for enhanced feature extraction and characterization of electromyogram signals | |
| CN110639169B (en) | CPM lower limb rehabilitation training method and system based on game and electromyographic signals | |
| CN105534517A (en) | Method for removing vehicle motion noise in three-lead electrocardiosignal | |
| CN111329476B (en) | Method and device for estimating muscle strength based on microscopic nerve driving information | |
| CN213758282U (en) | Surface electromyogram signal acquisition device | |
| CN110338791B (en) | Surface electromyogram signal active segment segmentation method based on sample entropy and Gaussian model | |
| CN105796091B (en) | A kind of intelligent terminal for removing electrocardiosignal vehicle movement noise | |
| Phinyomark et al. | Applications of variance fractal dimension: A survey | |
| CN112401905B (en) | Natural action electroencephalogram recognition method based on source localization and brain network | |
| CN110738093B (en) | Classification method based on improved small world echo state network electromyography | |
| CN110321856B (en) | Time-frequency multi-scale divergence CSP brain-computer interface method and device | |
| CN111839847A (en) | Myoelectric prosthesis control method, myoelectric prosthesis control device, electronic equipment and computer readable medium | |
| CN116628420A (en) | Brain wave signal processing method based on LSTM neural network element learning | |
| CN108836316B (en) | Electrocardiosignal R wave extraction method based on BP neural network | |
| Kocer | Classifying myopathy and neuropathy neuromuscular diseases using artificial neural networks | |
| Xue et al. | SEMG based intention identification of complex hand motion using nonlinear time series analysis | |
| Karlιk et al. | Differentiating types of muscle movements using a wavelet based fuzzy clustering neural network |
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 | ||
| CB02 | Change of applicant information |
Country or region after: China Address after: Room 306, building 4, future venture Plaza, hi tech Zone, Jinan City, Shandong Province Applicant after: Shandong Zhongke advanced technology Co.,Ltd. Address before: Room 306, building 4, future venture Plaza, hi tech Zone, Jinan City, Shandong Province Applicant before: Shandong Zhongke Advanced Technology Research Institute Co.,Ltd. Country or region before: China |
|
| CB02 | Change of applicant information |