CN116172584B - Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method - Google Patents
Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method Download PDFInfo
- Publication number
- CN116172584B CN116172584B CN202211608496.1A CN202211608496A CN116172584B CN 116172584 B CN116172584 B CN 116172584B CN 202211608496 A CN202211608496 A CN 202211608496A CN 116172584 B CN116172584 B CN 116172584B
- Authority
- CN
- China
- Prior art keywords
- channel
- signal
- len
- tke
- quality
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 31
- 238000001303 quality assessment method Methods 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 claims abstract description 22
- 230000003183 myoelectrical effect Effects 0.000 claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 238000013441 quality evaluation Methods 0.000 claims abstract description 11
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 230000003139 buffering effect Effects 0.000 claims abstract description 3
- 210000003205 muscle Anatomy 0.000 claims description 17
- 238000001228 spectrum Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 230000003044 adaptive effect Effects 0.000 claims 4
- 230000007774 longterm Effects 0.000 abstract description 5
- 238000004422 calculation algorithm Methods 0.000 description 22
- 230000009471 action Effects 0.000 description 11
- 230000008859 change Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 238000002567 electromyography Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000004118 muscle contraction Effects 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 206010021118 Hypotonia Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000036640 muscle relaxation Effects 0.000 description 1
- 230000002107 myocardial effect Effects 0.000 description 1
- 230000002232 neuromuscular Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 210000004243 sweat Anatomy 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/389—Electromyography [EMG]
- A61B5/397—Analysis of electromyograms
-
- 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
- A61B5/7221—Determining signal validity, reliability or quality
-
- 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
- A61B5/7253—Details of waveform analysis characterised by using transforms
- A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Heart & Thoracic Surgery (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Power Engineering (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
A self-adaptive multichannel electromyographic signal real-time quality evaluation and detection method comprises the following steps: step 1: buffering one window length of the original signal from the acquisition system at a time; step 2: performing normal form calibration; step 3: original signal x for all channels i (t) by w len The length is subjected to fast Fourier transform FFT, and then frequency domain energy occupancy sign FR is extracted i For subsequent channel quality assessment; step 4: original signal x for each channel in turn i (t) performing 20-500Hz band-pass filtering and 50Hz notch to remove part of noise component to obtain filtered signal y i (t); step 5: calculation of the myoelectric seizure period time duty cycle characteristic TR i The method comprises the steps of carrying out a first treatment on the surface of the Step 6: calculating the output value OUT of each channel i The method comprises the steps of carrying out a first treatment on the surface of the Step 7: adaptively updating a threshold value; step 8: evaluating signal quality Q of each channel i The method comprises the steps of carrying out a first treatment on the surface of the Step 9: according to channel quality factor Q i And calculating the weight of each channel to obtain the myoelectricity detection result at the step moment. The utility model has self-adaptive threshold calibration capability, and improves the reliability of long-term detection.
Description
Technical Field
The utility model relates to the field of electromyographic signal processing and detection, in particular to a real-time detection method based on electromyographic signals.
Background
The surface electromyographic signals are the electrical signals accompanied by muscle contraction, are important methods for noninvasively detecting muscle activity on the body surface, and can reflect the activity of neuromuscular to a certain extent. The surface electromyographic signals can be effectively detected, and the medical detection, the recovery of patients and the human body auxiliary preconditions which take the surface electromyographic signals as input sources are realized. The collection of the surface electromyographic signals is completed through the form of electrodes attached to the surface of the skin, however, the surface electromyographic signals collected by the method are often mixed with noise and interference.
Because the presence or absence of the surface electromyographic signal is specifically expressed on the time domain waveform of the acquired electrical signal, a certain reference value is usually required to be manually set, and when the characteristics extracted from the electrical signal are greater than the reference value, the surface electromyographic signal is considered to appear; otherwise, no surface electromyographic signals are present. In the Chinese patent specification CN115114962A, a surface electromyographic signal detection method is disclosed, wherein the method is based on that the amplitude of the collected electrical signal is directly compared with a signal amplitude reference value, and when the amplitude reference value is exceeded, a muscle contraction signal is considered to be collected; otherwise, no electromyographic signal exists. In another chinese patent specification CN110236538A, a method for detecting a surface electromyographic signal is also disclosed, where the method uses a TKE operator to perform feature extraction on an original collected signal, where the TKE operator is a nonlinear operator that responds to both amplitude and frequency, and when the amplitude and frequency of the original signal increase, the output result of the TKE operator also increases correspondingly. The method requires the subject to do N actions at random, each action is separated by 2 seconds, then a series of preprocessing such as filtering is carried out on the acquired signals, then a TKE operator is calculated, and then data of a plurality of channels are directly averaged to obtain the average TKE output; and finally, calculating a threshold value, and determining a final binary output according to the TKE threshold value and the signal amplitude.
In the first surface electromyographic signal detection algorithm, the frequency information in the signal is ignored because the acquired electrical signal is directly compared with the reference value. The surface electromyographic signals are reflected in the time domain and the frequency domain, and a part of information hidden in the signals is lost, so that the detection capability of the surface electromyographic signals is reduced under the condition of low signal-to-noise ratio. Secondly, the reference value of the electric signal needs to be manually set, and the reference value is usually different from person to person, so that a threshold is improved for the application of the algorithm in an actual scene, and the items needing manual intervention are increased.
In the second surface myoelectricity detection algorithm, the TKE operator is adopted, so that the problem that the signal frequency domain component in the first algorithm cannot be effectively utilized is solved to a certain extent. And secondly, the threshold value of the TKE operator is calculated according to the TKE output value calculated by the noise section signal, so that the step of manually setting a reference value is omitted, and the possibility of practical application of the algorithm is improved. However, this algorithm requires a 2 second interval between each action of the subject, does not account for the nature of automatic updating of the threshold, and increases the limitations of the algorithm. In addition, after the electromyographic signals of the subjects are acquired through a plurality of channels, the algorithm directly averages the data of each channel, and the robustness of the algorithm is lost. When a channel, or even multiple channels, has poor signal quality due to poor contact, the average result will introduce a significant degree of these disturbances into the system.
Further, since the acquired myoelectric signals are strongly correlated with the relative positions of the patch electrodes and the muscles, in the conventional myoelectric acquisition system, the user is required to carefully and correctly wear the myoelectric acquisition device. This increases the user's threshold of use.
In an actual application scene, when a user uses the surface myoelectric electrode at a muscle part, under the condition of wearing for a long time, the myoelectric signal and the state of noise can be slowly changed due to various reasons. For example:
(1) The electrode is relatively displaced due to muscle movement, so that the ideal fitting position of the electrode and the muscle is changed, and the surface myoelectricity detection result is affected.
(2) The change in skin condition (sweat, etc.) due to the wearing site causes a change in impedance, which in turn causes a change in signal.
(3) The change in the use environment of the user causes the baseband noise to change.
(4) The frequency and amplitude of the electromyographic signals change due to the muscles being in tension for a long period of time.
In addition to this, there are other factors that affect the electromyographic signal and noise. At this point the algorithm may not be able to effectively track such changes, resulting in reduced reliability of the algorithm.
The self-adaptive capacity of the existing technology for the change is relatively insufficient, and the algorithm with the signal state tracking capacity has higher requirement on hardware computing power, so that the real-time performance in the actual application scene can not be ensured.
On the other hand, when worn for a long time, the surface myoelectric electrode is easy to contact with skin poorly, so that the signal-to-noise ratio is reduced, and the motion artifact is large.
Disclosure of Invention
The utility model aims to overcome the problems in the prior art and provides a self-adaptive multichannel electromyographic signal real-time quality evaluation and detection method. The multi-channel acquisition device is suitable for multi-channel acquisition on the same muscle and detection of the existence of the electromyographic signals.
The utility model discloses a self-adaptive multichannel electromyographic signal real-time quality evaluation and detection method, which comprises the following steps:
step 1: buffering one window length W at a time from the acquisition system len Original signal x of (2) 1 (t)，x 2 (t)，...，x n (t), wherein 0.ltoreq.t < W len N is the number of channels; adjacent windows are spaced apart by a certain step S len ；
Step 2: performing normal form calibration; the normal calibration specifically includes:
step 2-1: the subject firstly releases muscle, at this time, collects the reference signal and a certain time length L 1 After that, the muscle is contracted according to the normal form prompt, and a certain time length L is acquired 2 Loosening again; the acquired reference signal comprises myoelectric rest potential and noise signal, b 1 (t)，b 2 (t)，...，b n (t)，0≤t＜L 1 ；
Step 2-2: according to the reference signal b i (t) calculating TKE output results to obtain initial TKE threshold Th of each channel i I is the current channel number; the TKE operator is specifically, for a sequence x (n), TKE output thereofThe calculation method comprises the following steps:
the calculation method of the TKE threshold value comprises the following steps:
Th＝μ+k*σ (2)
in the method, in the process of the utility model,mu isMean value of σ +.>Standard deviation of (2); k is a weighted value, and the default value of k is 6-8;
step 3: original signal x for all channels i (t) according to W len Length-fast fourier transform FFT, where x i (t) is a sampling signal of an ith channel, and t is a current point sequence number; then extracting frequency domain energy occupancy sign FR i For subsequent channel quality assessment, the calculation method is as follows:
FR i ＝(E total -E th )/E total (3)
wherein E is total For total energy of frequency spectrum E th Is energy with the frequency more than 60 Hz;
step 4: original signal x for each channel in turn i (t) performing 20-500Hz band-pass filtering and 50Hz notch to remove part of noise component to obtain filtered signal y i (t)；
Step 5: filtered signals y for the respective channels i (t) calculating its latest S len TKE output value of individual pointsThen point by point with TKE threshold Th of the channel i Comparing to obtain myoelectric seizure sequence T of each channel i (t)：
At the same time, T is obtained according to the above i (t) calculating the time duty cycle characteristic TR of the myoelectric seizure period i For subsequent signal quality assessment; TR (TR) i The calculation method of (2) is as follows:
i.e. electromyographic signal duration/total signal input duration;
step 6: calculating the output value OUT of each channel i Setting an output threshold Th d At a fixed value, if TR i ＞Th d Then consider that the current step length has an electromyographic signal, OUT i Taking 1; otherwise, it is regarded as noise signal, OUT i Taking 0. Namely:
step 7: adaptively updating a threshold value; if OUT i If=0, dynamically updating TKE threshold value by taking the data, and calculating average power of the signal, which is regarded as noise power P n The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, the TKE threshold is not updated, and the average power of the signal is calculated and taken as the electromyographic signal power P s ；
And calculates the signal to noise ratio according to the noise and the electromyographic signal power
Wherein P is n Is the noise power, P s The myoelectric signal power;
step 8: evaluating signal quality Q of each channel i The method comprises the steps of carrying out a first treatment on the surface of the According to signal-to-noise ratio SNR i Time domain duration of electromyographic signals is equal to TR i Electromyographic signal frequency domain energy duty cycle FR i Output channel quality factor
Q i ＝log(SNR i )+FR i +TR i (8)
The quality evaluation result of each channel is that
Quality i Is a slaveA score of 0% to 100% indicating to the user the quality of each channel of myoelectricity acquisition;
step 9: according to channel quality factor Q i Calculating weights of the channels
The sum of all channel weights adds up to 100%; output result OUT for each channel according to each channel weight i Weighting is carried out, and then the myoelectricity detection result at the step length moment is obtained by classification:
outputting 1 when more than 50% of weight exists and the electromyographic signal exists in the current step length; otherwise, the output is 0, namely no electromyographic signal.
Further, in step 1, adjacent windows are spaced apart by a certain step S len ，W len The value is greater than 1s, S len The value is less than 300ms.
Further, in step 2, if at L 2 And if myoelectricity acting potential cannot be detected in all channels within the length, triggering overtime alarm and prompting the user of failure in calibration, and re-wearing myoelectricity acquisition equipment for re-calibration is needed.
Further, in step 7, dynamically updating the TKE threshold specifically includes:
step 7-1: assuming an overall Z 1 ...Z m+n The number m+n, which comprises only two subgroups x 1 ...x n ，y 1 ...y m The first group is collected historical data, x 1 ...x n Mean and standard deviation of (2) are respectivelySum sigma x The second group is the new data currently acquired, y 1 ...y m Mean and standard deviation of +.>Sum sigma y Average value of the population->And standard deviation sigma z The following are provided:
thus, the current updated threshold value can be obtained by the historical data and the current data
Step 7-2: the statistics of a part of the subset are taken from the statistics of the whole data, and the following formula is referred to:
then fromThe updated threshold value is obtained.
Further, in step 8, when Quality i When the quality of the channel i is less than 30%, an alarm is sent to a user to remind the user that the quality of the channel i is extremely poor and the user needs to wear the channel i again.
The principle of the utility model is as follows: firstly, acquiring multichannel original myoelectric signals of target muscles, on one hand, finishing signal preprocessing and denoising through bandpass filtering and the like, then calculating TKE characteristics of each channel according to TKE operators, carrying out smoothing processing on the TKE characteristics, then judging the attack condition, and considering that the channel has myoelectric attack if the condition is met, wherein the initial threshold of the TKE operators is obtained through base noise calibration of a muscle relaxation state, and continuously and adaptively updating according to a statistical method in the following process. On the other hand, the time-frequency domain quality evaluation factors and the signal-to-noise ratio are extracted from the original electromyographic signals, and the real-time quality evaluation is carried out on each channel to generate the weighting weight of each channel. And finally, inputting the weight of each channel and the detection result of each channel into a classifier, and outputting the detection result of the final electromyographic signal.
The beneficial effects of the utility model are as follows:
first, the utility model has self-adaptive threshold calibration capability, and when a user wears the myoelectricity acquisition equipment for the first time, the myoelectricity threshold calibration can be realized by simple action, and then the threshold can be self-adaptively adjusted along with time, so that the reliability of long-term detection is improved.
And secondly, the quality of the channels is automatically evaluated and the weight is calculated among the channels, so that a user does not need to relate to the problem of poor contact between the optimal acquisition site and part of electrodes of myocardial electricity, the usability of the myoelectricity acquisition equipment is greatly improved, and meanwhile, the stability of myoelectricity long-term detection is improved.
Third, the algorithm requires very little hardware effort. Taking ESP32-S3 embedded chips as an example, through testing, the algorithm can run in real time in the embedded system, so that the timely response speed is ensured, and the algorithm is more practical to land.
In addition, the utility model has a great characteristic of strong expansibility. The method can be combined with muscle detection at a plurality of sites and subsequent feature extraction and action classifier to realize complex application scenes such as arm action recognition.
Drawings
Fig. 1 is a flowchart of the whole.
Fig. 2 is a flowchart of the feature extraction section and TKE threshold automatic updating section.
FIG. 3 is a flow chart of a channel quality assessment section.
Fig. 4 is a flow chart of a paradigm alignment part.
Fig. 5 is a flow chart of the classification section.
FIG. 6 is W len And S is equal to len Is a graph of the relationship of (1).
Detailed Description
In order to make the objects, technical solutions and advantages of the present utility model more apparent, the present algorithm will be described in further detail below with reference to the accompanying drawings and examples. It should be understood that the detailed description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the utility model.
The utility model discloses a self-adaptive multichannel electromyographic signal real-time quality evaluation and detection method, and an overall algorithm architecture is shown in figure 1, and can be divided into six sub-modules of data preprocessing, normal form calibration, feature extraction, self-adaptive updating, channel quality evaluation and detection output. Assuming that the presence or absence of an electromyographic signal is a switching value, the switching values are respectively expressed by 1 and 0; the number of the channels of the acquisition system is n, and the specific steps of the algorithm are as follows:
step one: the data preprocessing module caches one window length (W) at a time from the acquisition system len ) Original signal x of (2) 1 (t)，x 2 (t)，...，x n (t), wherein 0.ltoreq.t < W len N is the number of channels. Adjacent windows are separated by a certain step (S len )，W len And S is equal to len The relationship of (2) is shown in FIG. 6. In general, W len Can take a value of more than 1s, S len Can take a value less than 300ms
Step two: and performing normal form calibration. An example implementation of the paradigm calibration is: the subject firstly releases muscle, at this time, collects a reference signal, and collects a certain length L 1 Then, according to the normal form prompt (the normal form prompt refers to a series of actions designed in advance, the user can be prompted to do what action currently by means of voice, indicator lights or videos, and the like), the muscle is contracted, and a certain length L is acquired 2 And then relax.
The purpose of the model calibration is to acquire a reference signal b from the electrode acquisition when the muscle of the subject is relaxed 1 (t)，b 2 (t)，...，b n (t)，0≤t＜L 1 . Then according to the reference signal b of each channel i (t) calculation of TKE (Tea)A ger-Kaiser Energy operator) to obtain initial TKE threshold Th of each channel i Where i is the current channel number.
The TKE operator used is specifically, for a sequence x (n), its TKE outputThe calculation method comprises the following steps:
the calculation method of the TKE threshold value is as follows
Th＝μ+k*σ (2)
Wherein mu isMean value of σ +.>K is a weighted value, and the default value can be 6-8.
If in the process of calibration, if in L 2 And if myoelectricity acting potential cannot be detected in all channels within the length, triggering overtime alarm and prompting the user of failure in calibration, and re-wearing myoelectricity acquisition equipment for re-calibration is needed.
Step three: after calibration is completed, the subject arbitrarily makes actions, the duration of the actions and the time interval between the actions are arbitrary, and a signal x is acquired 1 (t)，x 2 (t)，...，x n (t). After acquisition, the acquisition system is S according to the step length len =256 samples, window length W len 2048 signal x i (t) output to the signal processing subsystem.
The original signal x of the signal processing subsystem for all channels i (t) according to W len The length is subjected to a Fast Fourier Transform (FFT) and then the frequency domain energy occupancy sign FR is extracted i For subsequent channel quality assessment, the calculation method is as follows:
FR i ＝(E total -E th )/E total (3)
wherein E is total For total energy of frequency spectrum E th Is energy with a frequency greater than 60 Hz. Because the effective range of the surface electromyographic signals in the frequency domain is between 10Hz and 500Hz, and the low-frequency noise and the power frequency interference are usually in the frequency band below 50Hz, the effect of distinguishing the low-frequency noise while extracting the surface electromyographic signals can be achieved by taking 60 frequency threshold values.
Step four: original signal x for each channel in turn i (t) performing 20-500Hz band-pass filtering and 50Hz notch to remove part of noise component to obtain filtered signal y i (t)。
Step five: filtered signals y for the respective channels i (t) calculating its latest S len TKE output value of individual pointsThen point by point with TKE threshold Th of the channel i Comparing to obtain myoelectric seizure sequence T of each channel i (t)：
At the same time, T is obtained according to the above i (t) calculating the time duty cycle characteristic TR of the myoelectric seizure period i For subsequent signal quality assessment.
I.e. electromyographic signal duration/total signal input duration.
Step six: calculating the output value OUT of each channel i . First, an output threshold Th is selected d Here Th d Can be set to a fixed value by human. The threshold value is set to Th d For example, =2%, if TR i ＞Th a Then consider that the current step length has an electromyographic signal, OUT i Taking 1; otherwise, it is regarded as noise signal, OUT i Taking 0. Namely:
step seven: the threshold is adaptively updated. If OUT i If=0, dynamically updating TKE threshold value by taking the data, and calculating average power of the signal, which is regarded as noise power P n The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, the TKE threshold is not updated, and the average power of the signal is calculated and taken as the electromyographic signal power P s . Calculating the signal to noise ratio according to the noise and the electromyographic signal power
The principle of dynamically updating the TKE threshold is as follows: assuming an overall Z 1 ...Z m+n The number m+n, which comprises only two subgroups x 1 ...x n ，y 1 ...y m The first group is collected historical data, x 1 ...x n Mean and standard deviation of (2) are respectivelySum sigma x The second group is the new data currently acquired, y 1 ...y m Mean and standard deviation of +.>Sum sigma y Average value of the population->And standard deviation sigma z The following are provided:
thus, the current updated threshold value can be obtained by the historical data and the current data
Next, to extract statistics of a subset from statistics of the whole data (here, to remove the effect of historical data over time on the threshold value), reference is made to the following formula:
then fromThe updated threshold value is obtained.
Step eight: evaluating signal quality Q of each channel i . According to signal-to-noise ratio SNR i Time domain duration of electromyographic signals is equal to TR i Electromyographic signal frequency domain energy duty cycle FR i Output channel quality factor
Q i ＝log(SNR i )+FR i +TR i (8)
The quality evaluation result of each channel is that
Quality i A score from 0% to 100% can be used to indicate to the user the quality of the myoelectric acquisition of each channel. For example, when Quality i When the quality of the channel i is less than 30%, an alarm is sent to a user to remind the user that the quality of the channel i is extremely poor and the user needs to wear the channel i again.
Step nine: according to channel quality factor Q i Calculation ofWeights of individual channels
The sum of all channel weights adds to 100%. Output result OUT for each channel according to each channel weight i Weighting is carried out, and then the myoelectricity detection result at the step length moment is obtained by classification:
outputting 1 when more than 50% of weight exists and the electromyographic signal exists in the current step length; otherwise, the output is 0, namely 'no electromyographic signal'.
Aiming at the problems that the self-adaptive capacity of the existing technology for influencing the changes of the electromyographic signals and the noise is relatively insufficient, and the algorithm with the signal state tracking capacity has higher requirement on hardware computing power and cannot guarantee the real-time performance in the practical application scene, the utility model derives the integral noise of the sample and the statistical distribution of the electromyographic signals from real-time data by a statistical method, derives the signal state in the last period according to the integral statistical distribution, and automatically updates the characteristic value obtained by the noise signal calculation. The method ensures that the algorithm can effectively track the slow change of the signal and ensures the real-time performance of the algorithm, thereby solving the technical problem.
Aiming at the problems of poor contact between the surface myoelectricity electrode and skin and high signal to noise ratio and motion artifact caused by long-term wearing in the prior art, the utility model automatically evaluates the signal quality of each channel and adjusts the weight of each channel in real time by arranging the multi-channel electrode on the same muscle, thereby improving the accuracy of myoelectricity detection and the stability of long-term wearing.
Meanwhile, the utility model solves the problems that in the surface electromyography signal detection technology based on amplitude, electromyography is needed to be observed manually, a signal section of an electromyography part is found, and the reference quantity of an algorithm is set manually, and the threshold used by a user is reduced to a great extent.
The embodiments described in the present specification are merely examples of implementation forms of the inventive concept, and the scope of protection of the present utility model should not be construed as being limited to the specific forms set forth in the embodiments, and the scope of protection of the present utility model and equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.
Claims (4)
1. The adaptive multichannel electromyographic signal real-time quality evaluation and detection method is characterized by comprising the following steps of:
step 1: buffering one window length W at a time from the acquisition system len Original signal x of (2) 1 (t),x 2 (t),…,x n (t), wherein 0.ltoreq.t<W len N is the number of channels; adjacent windows are spaced apart by a certain step S len ；
Step 2: performing normal form calibration; the normal calibration specifically includes:
step 2-1: the subject firstly releases muscle, at this time, collects the reference signal and a certain time length L 1 After that, the muscle is contracted according to the normal form prompt, and a certain time length L is acquired 2 Loosening again; acquiring a reference signal b 1 (t),b 2 (t),…,b n (t),0≤t<L 1 ；
Step 2-2: according to the reference signal b i (t) calculating TKE output results to obtain initial TKE threshold Th of each channel i I is the current channel number; the TKE operator is specifically, for a sequence x (n), TKE output thereofThe calculation method comprises the following steps:
the calculation method of the TKE threshold value comprises the following steps:
Th＝μ+k*σ (2)
wherein mu isMean value of σ +.>Standard deviation of (2); k is a weighted value, and the default value of k is 6-8;
step 3: original signal x for all channels i (t) according to W len Length-fast fourier transform FFT, where x i (t) is a sampling signal of an ith channel, and t is a current point sequence number; then extracting frequency domain energy occupancy sign FR i For subsequent channel quality assessment, the calculation method is as follows:
FR i ＝(E total －E th )/E total (3)
wherein E is total For total energy of frequency spectrum E th Is energy with the frequency more than 60 Hz;
step 4: original signal x for each channel in turn i (t) performing 20-500Hz band-pass filtering and 50Hz notch to remove part of noise component to obtain filtered signal y i (t)；
Step 5: filtered signals y for the respective channels i (t) calculating its latest S len TKE output value of individual pointsW len -S len ≤t<W len The method comprises the steps of carrying out a first treatment on the surface of the Then point by point with TKE threshold Th of the channel i Comparing to obtain myoelectric seizure sequence T of each channel i (t)：
At the same time, T is obtained according to the above i (t) calculating the time duty cycle characteristic TR of the myoelectric seizure period i For subsequent signal quality assessment; TR (TR) i The calculation method of (2) is as follows:
i.e. electromyographic signal duration/total signal input duration;
step 6: calculating the output value OUT of each channel i Setting an output threshold Th d At a fixed value, if TR i >Th d Then consider that the current step length has an electromyographic signal, OUT i Taking 1; otherwise, it is regarded as noise signal, OUT i Taking 0, namely:
step 7: adaptively updating a threshold value; if OUT i If=0, dynamically updating TKE threshold value by taking the data, and calculating average power of the signal, which is regarded as noise power P n The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, the TKE threshold is not updated, and the average power of the signal is calculated and taken as the electromyographic signal power P s ；
And calculates the signal to noise ratio according to the noise and the electromyographic signal power
Wherein P is n Is the noise power, P s The myoelectric signal power;
in step 7, dynamically updating the TKE threshold specifically includes:
step 7-1: assuming an overall Z 1 …Z m+n The number m+n, which comprises only two subgroups x 1 …x n ,y 1 …y m The first group is collected historical data, x 1 …x n Mean and standard deviation of (2) are respectivelySum sigma x The second group is the new data currently acquired, y 1 …y m Mean and standard deviation of +.>Sum sigma y Average value of the population->And standard deviation sigma z The following are provided:
thus, the current updated threshold value can be obtained by the historical data and the current data
Step 7-2: the statistics of a part of the subset are taken from the statistics of the whole data, and the following formula is referred to:
then fromThe updated threshold value can be obtained;
step 8: evaluating signal quality Q of each channel i The method comprises the steps of carrying out a first treatment on the surface of the According to signal-to-noise ratio SNR i Time domain duration of electromyographic signals is equal to TR i Electromyographic signal frequency domain energy duty cycle FR i Output channel quality factor
Q i ＝log(SNR i )+FR i +TR i (8)
The quality evaluation result of each channel is that
Quality i The score is from 0% to 100%, and the quality of each channel of myoelectricity acquisition is indicated to the user;
step 9: according to channel quality factor Q i Calculating weights of the channels
The sum of all channel weights adds up to 100%; output result OUT for each channel according to each channel weight i Weighting is carried out, and then the myoelectricity detection result at the step length moment is obtained by classification:
outputting 1 when more than 50% of weight exists and the electromyographic signal exists in the current step length; otherwise, the output is 0, namely no electromyographic signal.
2. The adaptive multi-channel electromyographic signal real-time quality assessment and detection method according to claim 1, wherein: in step 1, adjacent windows are separated by a certain step S len ，W len The value is greater than 1s, S len The value is less than 300ms.
3. The adaptive multi-channel electromyographic signal real-time quality assessment and detection method according to claim 1, wherein: in step 2, if at L 2 And if myoelectricity acting potential cannot be detected in all channels within the length, triggering overtime alarm and prompting the user of failure in calibration, and re-wearing myoelectricity acquisition equipment for re-calibration is needed.
4. The adaptive multi-channel electromyographic signal real-time quality assessment and detection method according to claim 1, wherein: in step 8, when Quality i <And when the number of the channel I is 30%, an alarm is sent to a user to remind the user that the quality of the channel I is extremely poor and the user needs to wear the channel I again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211608496.1A CN116172584B (en) | 2022-12-14 | 2022-12-14 | Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211608496.1A CN116172584B (en) | 2022-12-14 | 2022-12-14 | Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116172584A CN116172584A (en) | 2023-05-30 |
| CN116172584B true CN116172584B (en) | 2024-01-23 |
Family
ID=86447968
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211608496.1A Active CN116172584B (en) | 2022-12-14 | 2022-12-14 | Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116172584B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105943042A (en) * | 2016-06-07 | 2016-09-21 | 中国人民解放军国防科学技术大学 | Operator-hand-behavior perception system based on electromyographic signals |
| CN106308809A (en) * | 2016-08-15 | 2017-01-11 | 河北工业大学 | Method for recognizing gait of thigh amputation subject |
| KR20180026711A (en) * | 2018-03-05 | 2018-03-13 | 연세대학교 원주산학협력단 | Finger language recognition system and method |
| CN112221110A (en) * | 2020-06-03 | 2021-01-15 | 中国人民解放军陆军军医大学 | Beyond-obstacle movement wrist joint protection device and detection method |
-
2022
- 2022-12-14 CN CN202211608496.1A patent/CN116172584B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105943042A (en) * | 2016-06-07 | 2016-09-21 | 中国人民解放军国防科学技术大学 | Operator-hand-behavior perception system based on electromyographic signals |
| CN106308809A (en) * | 2016-08-15 | 2017-01-11 | 河北工业大学 | Method for recognizing gait of thigh amputation subject |
| KR20180026711A (en) * | 2018-03-05 | 2018-03-13 | 연세대학교 원주산학협력단 | Finger language recognition system and method |
| CN112221110A (en) * | 2020-06-03 | 2021-01-15 | 中国人民解放军陆军军医大学 | Beyond-obstacle movement wrist joint protection device and detection method |
Non-Patent Citations (1)
| Title |
|---|
| Hichem Bengacemi,等.A new detection method for EMG activity monitoring.Medical & Biological Engineering & Computing .2019,319–334. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116172584A (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Obeid et al. | Evaluation of spike-detection algorithms fora brain-machine interface application | |
| Faezipour et al. | A patient-adaptive profiling scheme for ECG beat classification | |
| CN101449973B (en) | Judgment index generation method and device for cardiac interference signal identification | |
| Fadlallah et al. | Weighted-permutation entropy: A complexity measure for time series incorporating amplitude information | |
| Nazarpour et al. | Application of higher order statistics to surface electromyogram signal classification | |
| CN108577834B (en) | A method of it is detected automatically for epilepsy interphase spike | |
| CN106691474A (en) | Brain electrical signal and physiological signal fused fatigue detection system | |
| CN107822617B (en) | Heartbeat anomaly detection method based on WiFi signals | |
| CN106485208B (en) | The automatic removal method of eye electrical interference in single channel EEG signals | |
| CN114052744B (en) | Electrocardiosignal classification method based on impulse neural network | |
| CN108888259A (en) | A kind of real-time QRS wave detection method of electrocardiosignal | |
| WO2023130600A1 (en) | Time domain data classification method based on zero crossing coefficient, and implantable stimulation system | |
| EP3206574A1 (en) | Frame based spike detection module | |
| CN109893115A (en) | A kind of processing analysis method based on human body weak biological electric signal | |
| CN114886388A (en) | Evaluation method and device for quality of electroencephalogram signal in anesthesia depth monitoring process | |
| Phinyomark et al. | EMG amplitude estimators based on probability distribution for muscle–computer interface | |
| CN110507317B (en) | Self-adaptive CA-CFAR (Carrier frequency-constant false alarm) positioning method for electrocardiosignal R wave | |
| Phinyomark et al. | EMG feature extraction for tolerance of 50 Hz interference | |
| CN117349598B (en) | Electroencephalogram signal processing method and device, equipment and storage medium | |
| CN116172584B (en) | Self-adaptive multichannel electromyographic signal real-time quality assessment and detection method | |
| Grönlund et al. | On-line signal quality estimation of multichannel surface electromyograms | |
| CN113378737A (en) | Implanted brain-computer interface neuron spike potential classification method | |
| CN110507299A (en) | Heart rate signal detection device and method | |
| CN105997067A (en) | Adaptive electromyographic signal detecting and processing method based on fractional Fourier transform | |
| CN114098768A (en) | Cross-individual surface electromyographic signal gesture recognition method based on dynamic threshold and easy TL |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |