CN113712541A - Rehabilitation method and system based on multi-sensor action recognition - Google Patents
Rehabilitation method and system based on multi-sensor action recognition Download PDFInfo
- Publication number
- CN113712541A CN113712541A CN202111031499.9A CN202111031499A CN113712541A CN 113712541 A CN113712541 A CN 113712541A CN 202111031499 A CN202111031499 A CN 202111031499A CN 113712541 A CN113712541 A CN 113712541A
- Authority
- CN
- China
- Prior art keywords
- rehabilitation
- sample data
- signal data
- action
- acceleration
- 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.)
- Granted
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1118—Determining activity level
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1121—Determining geometric values, e.g. centre of rotation or angular range of movement
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/389—Electromyography [EMG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—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]
- A61B5/395—Details of stimulation, e.g. nerve stimulation to elicit EMG response
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N1/00—Electrotherapy; Circuits therefor
- A61N1/18—Applying electric currents by contact electrodes
- A61N1/32—Applying electric currents by contact electrodes alternating or intermittent currents
- A61N1/36—Applying electric currents by contact electrodes alternating or intermittent currents for stimulation
- A61N1/36003—Applying electric currents by contact electrodes alternating or intermittent currents for stimulation of motor muscles, e.g. for walking assistance
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N1/00—Electrotherapy; Circuits therefor
- A61N1/18—Applying electric currents by contact electrodes
- A61N1/32—Applying electric currents by contact electrodes alternating or intermittent currents
- A61N1/36—Applying electric currents by contact electrodes alternating or intermittent currents for stimulation
- A61N1/36014—External stimulators, e.g. with patch electrodes
- A61N1/3603—Control systems
- A61N1/36031—Control systems using physiological parameters for adjustment
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B71/00—Games or sports accessories not covered in groups A63B1/00 - A63B69/00
- A63B71/06—Indicating or scoring devices for games or players, or for other sports activities
- A63B71/0619—Displays, user interfaces and indicating devices, specially adapted for sport equipment, e.g. display mounted on treadmills
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B71/00—Games or sports accessories not covered in groups A63B1/00 - A63B69/00
- A63B71/06—Indicating or scoring devices for games or players, or for other sports activities
- A63B71/0619—Displays, user interfaces and indicating devices, specially adapted for sport equipment, e.g. display mounted on treadmills
- A63B2071/0655—Tactile feedback
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2213/00—Exercising combined with therapy
- A63B2213/004—Exercising combined with therapy with electrotherapy
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2214/00—Training methods
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2220/00—Measuring of physical parameters relating to sporting activity
- A63B2220/40—Acceleration
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2230/00—Measuring physiological parameters of the user
- A63B2230/08—Measuring physiological parameters of the user other bio-electrical signals
- A63B2230/085—Measuring physiological parameters of the user other bio-electrical signals used as a control parameter for the apparatus
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Physics & Mathematics (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Pathology (AREA)
- Physiology (AREA)
- Artificial Intelligence (AREA)
- Physical Education & Sports Medicine (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Dentistry (AREA)
- Fuzzy Systems (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Human Computer Interaction (AREA)
- Psychiatry (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to a rehabilitation method and a rehabilitation system based on multi-sensor action recognition, which specifically comprise the following steps: sample data obtaining step: acquiring acceleration sample data of rehabilitation actions, and adding the acceleration sample data into a database; local feature acquisition: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features; and (3) action recognition: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized; a compensation step: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data. The invention adopts a multi-sensor data fusion algorithm for human body intention identification, uses a full connection layer to assemble all local features into complete one-dimensional features through a matrix again, and finally uses a soft-max layer to map the one-dimensional features into probabilities to complete classification.
Description
Technical Field
The invention relates to a rehabilitation method and a rehabilitation system based on multi-sensor action recognition.
Background
In recent years, patients with lower limb motor dysfunction caused by central nervous system diseases such as spinal injuries, cerebral apoplexy and the like have a rapidly increasing trend, and the health of human beings is seriously harmed. With the development of society and the improvement of medical treatment and living standard of people, the health of disabled people draws attention of the whole society. The weight-reducing walking training is one of the important means for the walking rehabilitation treatment of the patients with the diseases, and a large number of clinical studies prove the effectiveness of the weight-reducing walking training. The traditional rehabilitation treatment method is mainly characterized in that a nursing master assists a patient to carry out rehabilitation training, the rehabilitation training effect depends on the technical level and love of the nursing master, meanwhile, the number of the nursing masters is seriously insufficient, the training efficiency is low, the working strength is high, and the rehabilitation training efficiency of the patient is difficult to improve rapidly.
The defects and shortcomings of the prior art are as follows:
1. the existing rehabilitation depends on the use of mechanical exercise rehabilitation, the exercise training of the mechanical exercise rehabilitation is stiff, and the training action is simplified.
2. The existing motor rehabilitation lacks of evaluation on nerves in the rehabilitation process, and an effective rehabilitation scheme cannot be made;
3. currently, the mainstream human body action recognition is based on a visual sensor, and the recognition effect is easily interfered by external factors such as illumination, angle and the like; deployment costs are expensive.
4. The advantages of heterogeneous sensors can be complemented by multi-sensor data fusion, but the multi-sensor data fusion is always a difficult point of research.
Disclosure of Invention
The invention aims to solve the defects of the prior art and provides a rehabilitation method based on multi-sensor action recognition, which comprises the following steps:
sample data obtaining step: acquiring acceleration sample data of rehabilitation actions, and adding the acceleration sample data into a database;
local feature acquisition: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features;
and (3) action recognition: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized;
a compensation step: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
A rehabilitation system based on multi-sensor motion recognition, the rehabilitation system comprising:
a sample data obtaining unit: acquiring acceleration sample data of rehabilitation actions, and adding the acceleration sample data into a database;
a local feature acquisition unit: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features;
an action recognition unit: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized;
a compensation unit: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
The invention has the following advantages:
1. the rehabilitation robot is controlled through human body intention recognition, an active control mode that a patient participates in rehabilitation is achieved, rehabilitation actions can be achieved, and the patient can be assisted to do certain daily simple activities.
2. A multi-sensor data fusion algorithm used for human body intention recognition is used for carrying out classification training on data of different sensors by using a CART decision tree respectively to obtain a plurality of local features, all the local features are assembled into a complete one-dimensional feature through a matrix again by using a full connection layer (FC), and finally the one-dimensional feature is mapped into probability by using a soft-max layer to complete classification.
The above-described and other features, aspects, and advantages of the present application will become more apparent with reference to the following detailed description.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
fig. 1 is a schematic diagram of a rehabilitation method based on multi-sensor motion recognition.
Fig. 2 is a schematic diagram of softMax layer of a rehabilitation method based on multi-sensor action recognition.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the invention without any inventive step, are within the scope of protection of the invention.
Unless defined otherwise, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The use of "first," "second," and similar terms in the description and claims of the present application do not denote any order, quantity, or importance, but rather the terms are used to distinguish one element from another. Also, the use of the terms "a" or "an" and the like do not denote a limitation of quantity, but rather denote the presence of at least one.
A rehabilitation method based on multi-sensor action recognition comprises the following steps:
sample data obtaining step: acquiring acceleration sample data of the rehabilitation action, and adding the acceleration sample data into a database;
local feature acquisition: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features;
the invention adopts CART decision tree for classification training, and the classification tree (decision tree) is a very common classification method. It is a kind of supervised learning, which is to say, given a pile of samples, each sample has a set of attributes and a class, which are determined in advance, a classifier is obtained through learning, and the classifier can give correct classification to newly appeared objects. Such machine learning is called supervised learning.
The action recognition step specifically comprises the following steps: synchronously acquiring acceleration signal data and surface electromyogram signal data; carrying out feature extraction on the surface electromyographic signal data to obtain surface electromyographic single-dimensional features; judging whether a signal is triggered or not according to the surface electromyographic signal data characteristics, and if the signal is triggered, starting to extract the multidimensional local characteristics of the acceleration signal data according to a plurality of local characteristics; a full connection layer is adopted to assemble the multidimensional local features and the surface myoelectricity single-dimensional features into complete one-dimensional features through a matrix; and mapping the one-dimensional features into the probability of rehabilitation actions according to the soft-max layer, so as to realize the identification of the actions.
As shown in fig. 1, after the patient wears the upper limb rehabilitation robot, the data acquisition system acquires surface myoelectricity and acceleration data during normal limb rehabilitation training and daily actions of the patient, and the data processing includes intercepting acceleration signals according to surface myoelectricity activity sections, smoothing data and aligning. The intention identification method is carried out by combining inertial sensor data and surface electromyography data: the CART decision tree is used as a base classifier. As shown in fig. 2, the full link layer and the softMax layer are used to perform decorrelation and recombination on the multidimensional local features to obtain one-dimensional features, and finally the softMax is used to map the one-dimensional features into probabilities to complete action classification and prediction.
A compensation step: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
The invention constructs a 'surface myoelectricity-action' mapping. Taking the muscle as a reference basis, muscle electricity compensation is carried out on the muscles of the limbs with dyskinesia through a functional electrical stimulation module (FES), and the affected limbs are helped to complete normal rehabilitation training and daily actions.
As an alternative, the rehabilitation method further comprises a sample data updating step, wherein the sample data updating step comprises the following steps: and calculating a second derivative of the probability change according to the probabilities of the rehabilitation actions obtained in the action identification step, if the second derivative is lower than a preset threshold, putting the acceleration signal data corresponding to the rehabilitation actions into a database, and if the second derivative is higher than or equal to the preset threshold, no action exists.
As an alternative, the rehabilitation method further comprises a sample data updating step, wherein the sample data updating step comprises the following steps: if the proportion of the sample data volume of a certain rehabilitation action in the acceleration sample data to the total sample data volume is lower than a preset threshold, calculating a second derivative of probability change according to the probability of the rehabilitation action obtained in the action identification step, if the second derivative is lower than the preset threshold, putting the acceleration signal data corresponding to the rehabilitation action into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
As an alternative, the rehabilitation method further comprises a sample data deleting step, wherein the sample data deleting step comprises the following steps: and if the probability obtained in the action recognition step of certain acceleration signal data is lower than a preset threshold value, deleting acceleration sample data close to the acceleration signal data in the database.
The identification precision of the invention depends on the sample data in the database, and the invention realizes the update and deletion of the sample data through the probability of the rehabilitation action. In addition, the problem that the identification precision of the CART decision tree is easy to appear is solved through observing the sample size.
A rehabilitation system based on multi-sensor motion recognition, the rehabilitation system comprising: a sample data obtaining unit: acquiring acceleration sample data of the rehabilitation action, and adding the acceleration sample data into a database; a local feature acquisition unit: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features; an action recognition unit: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized; a compensation unit: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
The specific realized functions of the action recognition unit are as follows: synchronously acquiring acceleration signal data and surface electromyographic signal data to perform feature extraction on the surface electromyographic signal data to obtain surface electromyographic single-dimensional features; judging whether a signal is triggered or not according to the surface electromyographic signal data characteristics, and if the signal is triggered, starting to extract the multidimensional local characteristics of the acceleration signal data according to a plurality of local characteristics; a full connection layer is adopted to assemble the multidimensional local features and the surface myoelectricity single-dimensional features into complete one-dimensional features through a matrix; and mapping the one-dimensional features into the probability of rehabilitation actions according to the soft-max layer, so as to realize the identification of the actions.
The rehabilitation system further comprises a sample data updating unit: and calculating a second derivative of the probability change according to the probabilities of the rehabilitation actions obtained in the action identification unit, if the second derivative is lower than a preset threshold, putting the acceleration signal data corresponding to the rehabilitation actions into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
The rehabilitation system further comprises a sample data updating unit: if the ratio of the sample data volume of a certain rehabilitation action in the acceleration sample data to the total sample data volume is lower than a preset threshold, calculating a second derivative of probability change according to the probability of the rehabilitation action obtained in the action recognition unit, if the second derivative is lower than the preset threshold, putting the acceleration signal data corresponding to the rehabilitation action into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
The rehabilitation system further comprises a sample data deleting unit: and if the probability obtained in the action recognition step of certain acceleration signal data is lower than a preset threshold value, deleting acceleration sample data close to the acceleration signal data in the database.
1. Different from a non-wearable exoskeleton rehabilitation robot commonly used in the market, the wearable exoskeleton rehabilitation robot is suitable for a wearable soft robot, the soft material can be more suitable for limbs, the limbs of a patient are effectively protected, and the occurrence of secondary injury is reduced. And the wearable soft robot is not driven by a motor, but helps the muscles to contract through Functional Electrical Stimulation (FES) to complete the action, and has small weight and wide application range.
1. The invention solves the problem that the traditional rehabilitation robot has a single passive rehabilitation mode, has both passive control and active control, enables a patient to actively participate in actions through surface myoelectric feedback and upper limb action prediction, and assists the patient to complete rehabilitation training and daily life.
2. The invention needs human intention identification for active control, which is different from the common video-based human action identification, and uses a non-visual sensor: the human body intention is predicted by fusing data of multiple sensors through an inertial sensor and surface myoelectricity.
3. The invention adopts multi-sensor data fusion which is always a difficult point of man-machine interaction, and the invention carries out decorrelation characteristic fusion on inertial sensor data and surface electromyographic data through a full connection layer and a soft-Max layer, thereby achieving the purpose of data fusion processing.
While there have been shown and described what are at present considered the fundamental principles and essential features of the invention and its advantages, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and the description is given here only for clarity, and those skilled in the art should integrate the description, and the embodiments may be combined appropriately to form other embodiments understood by those skilled in the art.
Claims (10)
1. A rehabilitation method based on multi-sensor action recognition is characterized by comprising the following steps:
sample data obtaining step: acquiring acceleration sample data of rehabilitation actions, and adding the acceleration sample data into a database;
local feature acquisition: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features;
and (3) action recognition: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized;
a compensation step: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
2. The rehabilitation method based on multi-sensor motion recognition according to claim 1, wherein the motion recognition step specifically comprises the following steps:
synchronously acquiring acceleration signal data and surface electromyogram signal data;
carrying out feature extraction on the surface electromyographic signal data to obtain surface electromyographic single-dimensional features;
judging whether a signal is triggered or not according to the surface electromyographic signal data characteristics, and if the signal is triggered, starting to extract the multidimensional local characteristics of the acceleration signal data according to a plurality of local characteristics;
a full connection layer is adopted to assemble the multidimensional local features and the surface myoelectricity single-dimensional features into complete one-dimensional features through a matrix;
and mapping the one-dimensional features into the probability of rehabilitation actions according to the soft-max layer, so as to realize the identification of the actions.
3. The rehabilitation method based on multi-sensor motion recognition according to claim 1, further comprising a sample data updating step, wherein the sample data updating step comprises the following steps:
and calculating a second derivative of the probability change according to the probabilities of the rehabilitation actions obtained in the action identification step, if the second derivative is lower than a preset threshold, putting the acceleration signal data corresponding to the rehabilitation actions into a database, and if the second derivative is higher than or equal to the preset threshold, no action exists.
4. The rehabilitation method based on multi-sensor motion recognition according to claim 1, further comprising a sample data updating step, wherein the sample data updating step comprises the following steps:
if the proportion of the sample data volume of a certain rehabilitation action in the acceleration sample data to the total sample data volume is lower than a preset threshold, calculating a second derivative of probability change according to the probability of the rehabilitation action obtained in the action identification step, if the second derivative is lower than the preset threshold, putting the acceleration signal data corresponding to the rehabilitation action into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
5. The rehabilitation method based on multi-sensor motion recognition according to claim 1, further comprising a sample data deleting step, wherein the sample data deleting step comprises the following steps:
and if the probability obtained in the action recognition step of certain acceleration signal data is lower than a preset threshold value, deleting acceleration sample data which are close to the acceleration signal data in the database.
6. A rehabilitation system based on multi-sensor motion recognition, the rehabilitation system comprising:
a sample data obtaining unit: acquiring acceleration sample data of rehabilitation actions, and adding the acceleration sample data into a database;
a local feature acquisition unit: according to acceleration sample data in a database, carrying out classification training by adopting a CART decision tree to obtain a plurality of local features;
an action recognition unit: according to the acceleration signal data and the surface electromyographic signal data which are synchronously acquired, and according to the plurality of local characteristics and the single-dimensional characteristics of the surface electromyographic signal data, the action recognition is realized;
a compensation unit: myoelectric compensation is performed based on the identified motion and surface myoelectric signal data.
7. The rehabilitation system based on multi-sensor motion recognition according to claim 6, wherein the motion recognition unit specifically implements the following functions:
synchronously acquiring acceleration signal data and surface electromyogram signal data;
carrying out feature extraction on the surface electromyographic signal data to obtain surface electromyographic single-dimensional features;
judging whether a signal is triggered or not according to the surface electromyographic signal data characteristics, and if the signal is triggered, starting to extract the multidimensional local characteristics of the acceleration signal data according to a plurality of local characteristics;
a full connection layer is adopted to assemble the multidimensional local features and the surface myoelectricity single-dimensional features into complete one-dimensional features through a matrix;
and mapping the one-dimensional features into the probability of rehabilitation actions according to the soft-max layer, so as to realize the identification of the actions.
8. The multi-sensor motion recognition-based rehabilitation system according to claim 6, further comprising a sample data updating unit:
and calculating a second derivative of the probability change according to the probabilities of the rehabilitation actions obtained in the action identification unit, if the second derivative is lower than a preset threshold, putting the acceleration signal data corresponding to the rehabilitation actions into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
9. The multi-sensor motion recognition-based rehabilitation system according to claim 6, further comprising a sample data updating unit:
if the ratio of the sample data volume of a certain rehabilitation action in the acceleration sample data to the total sample data volume is lower than a preset threshold, calculating a second derivative of probability change according to the probability of the rehabilitation action obtained in the action recognition unit, if the second derivative is lower than the preset threshold, putting the acceleration signal data corresponding to the rehabilitation action into a database, and if the second derivative is higher than or equal to the preset threshold, no action is performed.
10. The rehabilitation system according to claim 6, further comprising a sample data deleting unit:
and if the probability obtained in the action recognition step of certain acceleration signal data is lower than a preset threshold value, deleting acceleration sample data which are close to the acceleration signal data in the database.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111031499.9A CN113712541B (en) | 2021-09-03 | 2021-09-03 | Rehabilitation method and system based on multi-sensor action recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111031499.9A CN113712541B (en) | 2021-09-03 | 2021-09-03 | Rehabilitation method and system based on multi-sensor action recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113712541A true CN113712541A (en) | 2021-11-30 |
| CN113712541B CN113712541B (en) | 2022-08-02 |
Family
ID=78681405
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111031499.9A Active CN113712541B (en) | 2021-09-03 | 2021-09-03 | Rehabilitation method and system based on multi-sensor action recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113712541B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106176134A (en) * | 2016-07-12 | 2016-12-07 | 王春宝 | A kind of method and system of autonomous training |
| CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人（合肥）国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
| CN111388863A (en) * | 2020-03-16 | 2020-07-10 | 合肥工业大学 | Limb electrical stimulation training system based on joint angle feedback |
| CN111950460A (en) * | 2020-08-13 | 2020-11-17 | 电子科技大学 | Muscle strength self-adaptive stroke patient hand rehabilitation training action recognition method |
| CN113261981A (en) * | 2021-05-21 | 2021-08-17 | 华南理工大学 | Quantitative assessment method and system for upper limb spasm based on surface myoelectric signal |
-
2021
- 2021-09-03 CN CN202111031499.9A patent/CN113712541B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106176134A (en) * | 2016-07-12 | 2016-12-07 | 王春宝 | A kind of method and system of autonomous training |
| CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人（合肥）国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
| CN111388863A (en) * | 2020-03-16 | 2020-07-10 | 合肥工业大学 | Limb electrical stimulation training system based on joint angle feedback |
| CN111950460A (en) * | 2020-08-13 | 2020-11-17 | 电子科技大学 | Muscle strength self-adaptive stroke patient hand rehabilitation training action recognition method |
| CN113261981A (en) * | 2021-05-21 | 2021-08-17 | 华南理工大学 | Quantitative assessment method and system for upper limb spasm based on surface myoelectric signal |
Non-Patent Citations (1)
| Title |
|---|
| 唐广阳，郭建富: "《高等数学（上册）》", 30 April 2008, pages: 53 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113712541B (en) | 2022-08-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fang et al. | EMG-centered multisensory based technologies for pattern recognition in rehabilitation: state of the art and challenges | |
| CN103212156B (en) | The autonomous instrument of hemiplegic patient based on body image theory | |
| Samuel et al. | Pattern recognition of electromyography signals based on novel time domain features for amputees' limb motion classification | |
| Ewert et al. | What is Neuroethology? | |
| CN109366508A (en) | A kind of advanced machine arm control system and its implementation based on BCI | |
| CN203043423U (en) | Rehabilitation training device based on brain-computer interface | |
| CN111696645A (en) | Hand exoskeleton rehabilitation training device and method based on surface electromyographic signals | |
| CN202223388U (en) | Wearable brain-controlled intelligent prosthesis | |
| CN102309365A (en) | Wearable brain-control intelligent prosthesis | |
| CN106422203A (en) | Upper limb rehabilitation training method based on photoelectric multimode feedback of mirror image therapy | |
| Li et al. | EEG signal classification method based on feature priority analysis and CNN | |
| JP2009531077A (en) | Apparatus and method for real time control of effectors | |
| Bouton | Merging brain-computer interface and functional electrical stimulation technologies for movement restoration | |
| Li et al. | Human posture detection method based on wearable devices | |
| Guan et al. | Trunk muscle activity patterns and motion patterns of patients with motor complete spinal cord injury at T8 and T10 walking with different un-powered exoskeletons | |
| CN110908506B (en) | Bionic intelligent algorithm-driven active and passive integrated rehabilitation method, device, storage medium and equipment | |
| CN104825165B (en) | A kind of intelligent wearable device based on Ubiquitous Network transmission | |
| Merante et al. | Brain–Computer interfaces for spinal cord injury rehabilitation | |
| Kilgore et al. | Neuroprosthesis for individuals with spinal cord injury | |
| TW201831161A (en) | Lower limb rehabilitation system | |
| CN113712541B (en) | Rehabilitation method and system based on multi-sensor action recognition | |
| Zhang et al. | Novel approach for electromyography-controlled prostheses based on facial action | |
| CN209253488U (en) | A kind of bionical class brain intelligent hand electric mechanical ectoskeleton and its control system entirely | |
| Thomas et al. | Tactile unit properties after human cervical spinal cord injury | |
| Lou et al. | Wireless master-slave FES rehabilitation system using sEMG control |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230103 Address after: 201318 floors 4 and 5, building 1, No. 555, Qingdai Road, Pudong New Area, Shanghai Patentee after: SHANGHAI YINGSHOU MEDICAL INSTR Co.,Ltd. Address before: 200245 Nansha Road, Minhang District, Shanghai 68 Patentee before: NCC MEDICAL CO.,LTD. |
|
| TR01 | Transfer of patent right |