CN107918492A - A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods - Google Patents
A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods Download PDFInfo
- Publication number
- CN107918492A CN107918492A CN201711404559.0A CN201711404559A CN107918492A CN 107918492 A CN107918492 A CN 107918492A CN 201711404559 A CN201711404559 A CN 201711404559A CN 107918492 A CN107918492 A CN 107918492A
- Authority
- CN
- China
- Prior art keywords
- data
- conversion
- intention
- artificial limb
- intention assessment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/03—Arrangements for converting the position or the displacement of a member into a coded form
- G06F3/033—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor
- G06F3/0346—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor with detection of the device orientation or free movement in a 3D space, e.g. 3D mice, 6-DOF [six degrees of freedom] pointers using gyroscopes, accelerometers or tilt-sensors
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a kind of human motion intension recognizing method based on Intelligent lower limb artificial limb, the present invention solves the problems, such as that the recognition methods of conventional body's motion intention has hysteresis quality, before the conversion of the motor pattern of Ipsilateral of artificial limb is worn, according to embedded artificial limb or the time series data for the sensor generation for being bundled in strong side, the motion intention of disabled people patient is identified exactly, and, compared to multiple sensors used in traditional intension recognizing method, the present invention has only used a kind of sensor, that is inertial sensor, it becomes possible to which accurately identification human motion is intended to.The present invention comprises the following steps：Intention assessment data acquisition and Database；Intention assessment data prediction；Intention assessment data characteristics is extracted；Disaggregated model selects and model training and completion intention assessment.The present invention provides experimental data what recognition methods for the motion intention recognition methods of Intelligent lower limb artificial limb, and the development of intelligent artificial limb is promoted with this, preferably serves lower limb amputation patient.
Description
Technical field
The invention belongs to artificial intelligence and CRT technology field, and in particular to a kind of in face of Intelligent lower limb artificial limb
Human motion is intended to precognition recognition methods.
Background technology
For national Disability Sampling Survey the results show that the existing all kinds of disabled person's sums in China exceed 80,000,000, wherein limbs are residual
Disease person has seriously affected the normal life and work of individuals with disabilities more than the incompleteness of 24,000,000 limbs.Disability rehabilitation's artificial limb is limb
Body disabled person solves one of important means of action obstacle.Upper extremity prosthesis is different from, artificial leg control is related to human motion and puts down
Weighing apparatus problem, the influence to disabled person's daily life are very crucial.Although having had some commercialized artificial legs at present, its
The joint of middle overwhelming majority artificial limb is motorless.Disabled person dresses this kind of prosthetic walking, and 20%- is expended than Healthy People more
30% energy, for more complicated walking environment, such as stair and rough road surface, disabled person walks can be very
Painstakingly, and it is unable to maintain that kinetic stability.Realize that the Intelligent lower limb for introducing robot technology becomes international research hot spot, emphasis
It is main to include two aspects：The design of intelligent limb is with control and the human motion intention assessment based on inertial sensor is ground
Study carefully.The former is primarily upon how utilizing the mechanical structure and control method of intelligent bionic Technology design artificial limb, joint of artificial limb is existed
There is the mechanical characteristic closer to human synovial in walking process, and the latter is then concerned with how to be believed according to the human-body biological of collection
Number (e.g., surface electromyogram signal) and artificial limb sensor signal (e.g., acceleration and joint angles) identify the motion intention of people, and
The control parameter of artificial limb is adjusted according to recognition result, to realize the walking of nature, smoothness, stabilization.
In the control research of lower limb intelligent artificial limb, more commonly used control strategy is muti-layer control tactics.As shown in Fig. 2,
High level controller identifies the motion intention of people, and motion intention is converted to corresponding control algolithm, bottom controller by middle level controller
Closed-loop control, driving artificial limb movement are realized according to control algolithm.Motion intention recognition methods proposed by the invention belongs to high-rise
Control section.Human motion intention assessment plays a crucial role in intelligent power artificial leg control system is dressed,
The final purpose of human motion intention assessment is accurately and timely to decode the information of motion intention in people's nerve center, under intelligence
The bottom controller of limb artificial limb selects corresponding control strategy according to this motion intention information.Therefore, establish one it is perfect
Human motion intention assessment database towards Intelligent lower limb artificial limb is very necessary.
Researcher attempts various methods and causes database to include abundant artificial leg motion intention information, but in reality
Under the conditions of, establishing database, often there are problems with：1. experimental site is limited, the data scene of acquisition is single；2. disabled person
Volunteer's negligible amounts, and handicapped, make it wear artificial limb needs of normally walking and take considerable time and monetary cost.This hair
It is bright to attempt to evade problem above by rational database building method.
The content of the invention
To solve the above-mentioned problems, the present invention provides a kind of human motion in face of Intelligent lower limb artificial limb and is intended to predict identification
Method, the method is by establishing intention assessment database and data are embedded in artificial limb or being bundled in strong side, when wearing artificial limb
Before the motor pattern conversion of Ipsilateral, according to embedded artificial limb or the time series data for the sensor generation for being bundled in strong side, to deformity
The motion intention of people patient carries out precognition identification exactly；
Further, the described method includes：
S1：Intention assessment data acquisition simultaneously establishes database:Subject dresses inertial sensor, according to what is pre-defined
Take a step sequential movements, collect its exercise data and establish database；
S2：Intention assessment data prediction：The data collected are carried out with denoising, and abnormal data removes and repairing；
S3：Intention assessment data characteristics is extracted：Selection can distinguish different classes of motion characteristics attribute, and extract sample number
According to feature；
S4：Disaggregated model selects and model training, selects suitable disaggregated model to be trained；
S5：Intention assessment is completed, successfully identifies motion intention；
Further, the S1 is specifically included：
S11：2n subjects are selected, male to female ratio is equal, and wherein n is positive integer, and the subject of each health is simulating
With dressing inertial sensor respectively at shank and ankle, the inertial sensor is able to record for Ipsilateral and the thigh of the strong side of simulation
The acceleration and angular speed kinematics information at each moment during human motion, and with the sequential of time cumulation generation multiple data channel
Data, data preserve hereof frame by frame according to the structure of sensor；
S12：Alternating according to pre-defining is taken a step sequentially, simulates disabled person's stable state gait behavior and conversion gait successively
Behavior, the stable state gait behavior include level walking, and upstairs, downstairs, ascents and descents, the conversion gait behavior includes flat
Ground walk to go up a slope conversion, level walking to descending change, level walking to upstairs conversion, level walking to downstairs conversion, on
The conversion of slope aspect level walking, the conversion of lower slope aspect level walking, change to level walking and downstairs changed to level walking upstairs；
S13：The intention assessment data produced to inertial sensor are recorded and sorted out, and are converted into by data to be divided
The data file that analysis software is directly read, the recorded separately preservation of intention assessment data of each subject, and each different step
State behavior needs to put on corresponding label, the study and classification for carrying out having supervision for follow-up intention assessment algorithm；
Further, the S2 is specifically included：
S21：Denoising is carried out to initial data, abnormal data is rejected, missing data is repaired；
S22：With each step since liftoff to contacting to earth, long sequence is first cut into one small step of a small step, is then existed again
Adding window cuts into the window of 45 frames inside one small step；
S23：Windowing process is carried out to the data after cutting, makes all sample data window sizes equal, in the sample
With data be 45 frame data since less touch with the ground backward, the window of all data is all equal, and length of window is chosen
Be 45 frames, length is minimum, and every 45 frame is 1/4 gait cycle, the time be since a less touch with the ground, swing
For a period of time, tiptoe does not contact to earth also, the windowing process be specially so that all data be all since less touch with the ground backward
45 frame data；
S24：Required data are extracted in data in the window after well cutting again；
Further, for the sample data of conversion behavior, extraction conversion step in S24, the conversion step is in the past
One pattern is transformed into latter mode, the process that the same side less touch with the ground to foot contacts to earth；For stable state behavior, extract corresponding
Same foot is since liftoff next time to the data contacted to earth；
Further, the S3 is specifically included：
S31：For conversion behavior, the data in that leg of swing phase when extraction conversion step, for stable state behavior,
Extract the identical data of the swing phase of same foot, it is ensured that all data are consistent in sensing station, length of window；
S32：During extracting data characteristics, two legs alternately step, and in a gait cycle, every leg all undergoes two
In a stage, be respectively driving phase and recovery phase, wherein a leg is in driving phase, another leg is in recovery phase,
In the data of S24 well cuttings, three thigh, shank and ankle sensors in shaking peroid are chosen, and extract three respectively
Average, variance and the extreme value for the data that a sensor collects；
Further, the S4 includes：
S41：Selection support vector machines is classified；
S42：The SVM using radial basis function as kernel function is used, the feature extracted to intention assessment data is divided
Class；
S43：With reference to grid data service and K folding cross-validation method choose S42 in SVM optimized parameter, prevent over-fitting or
The generation of poor fitting phenomenon, optimal disaggregated model is obtained with this.
8th, the method according to the description of claim 7 is characterized in that SVM method for solving is as follows in S43：
s.t y(ω·xi+b)≥1-ξi, i=1,2,3 ..., n
ξ >=0,
Wherein, ξiReferred to as slack variable, C are penalty, and the form of Radial basis kernel function is as follows：
Wherein, σ is nuclear radius；
Further, it is right side that strong side is simulated in the S11, and simulation Ipsilateral is left side；
Further, motion intention can be identified in the S31 when motor pattern is changed, intelligent artificial limb
High level controller can change corresponding control parameter in advance with this result, better control over artificial limb；
Beneficial effects of the present invention are as follows：
1) before the conversion of the motor pattern of Ipsilateral of artificial limb is worn, according to embedded artificial limb or it is bundled in the sensor of strong side
The time series data of generation, identifies the motion intention of disabled people patient exactly, solves the knowledge of conventional body's motion intention
Other method has the problem of hysteresis quality；
2) comprising abundant artificial leg motion intention information in system, solving the experimental site under physical condition has
Limit, the data scene of acquisition are single；Volunteer with disabilities's negligible amounts, and it is handicapped, its wearing artificial limb is normally walked needs
Take considerable time and the problems such as monetary cost.
Brief description of the drawings
Fig. 1 is the flow chart of the method for the invention；
Fig. 2 is muti-layer control tactics figure described in background of invention；
Fig. 3 is windowing process schematic diagram in the method for the invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, it is right
The present invention is explained in further detail.It should be appreciated that specific embodiment described herein is used only for explaining the present invention, and
It is not used in the restriction present invention.On the contrary, the present invention cover it is any be defined by the claims the present invention spirit and scope on do
Replacement, modification, equivalent method and scheme.Further, in order to make the public have a better understanding the present invention, below to this
It is detailed to describe some specific detail sections in the detailed description of invention.It is thin without these for a person skilled in the art
The description of section part can also understand the present invention completely.
The invention will be further described with specific embodiment below in conjunction with the accompanying drawings, but not as a limitation of the invention.
Below most preferred embodiment is enumerated for the present invention：
As shown in Figure 1-Figure 3, the present invention provides a kind of human motion in face of Intelligent lower limb artificial limb and is intended to predict identification side
Method, the method implementation steps are as follows：
1)：Intention assessment data acquisition and Database, subject dress the biography of the mechanical informations such as good measurement acceleration
Sensor, according to the sequential movements of taking a step pre-defined, collects its exercise data and establishes database.Specifically：
Each subject the strong side of simulation (realized for convenience of description with algorithm, the strong side of regulation simulation is right side in the present invention,
Simulation Ipsilateral is left side) thigh and shank at dressed inertial sensor, inertial sensor is able to record every during human motion
The kinematics informations such as the acceleration at a moment, angular speed, and produce with time cumulation the time series data of multiple data channel.
Alternating according to pre-defining is taken a step sequentially, is simulated disabled person's level walking successively, is above gone downstairs, climb and fall, with
And the multiclass gait behavior such as level walking is gone downstairs upwards, climb and fall conversion.When establishing database, the present invention selects closes both at home and abroad
The common gait behavior in the correlative study of intension recognizing method, including：Level walking, upstairs, downstairs, goes up a slope, descending totally 5
Class stable state gait behavior and level walking are to conversion of going up a slope, level walking is changed to descending, level walking to change upstairs, it is flat
Ground walk to downstairs conversion, upper slope aspect level walking conversion, lower slope aspect level walking conversion, upstairs to level walking conversion, under
Gait behavior, totally 13 class gait behavior are changed to 8 classes such as level walking conversions in building.
In order to facilitate the processing of experimental data, in data acquisition, according to different behavior classifications, preset and feel better
The alternating of examination person is taken a step sequentially.For level walking, upstairs, downstairs, go up a slope, this 5 class stable state behavior of descending, the alternating taken steps
Step order be：Side-simulation Ipsilateral is good in the strong side-simulation Ipsilateral-simulation of simulation, or simulation Ipsilateral-simulation is good for side-simulation and is suffered from
Side-simulation is good for side and is replaced successively.For level walking to conversion of going up a slope, level walking to descending conversion, level walking to upstairs
Conversion, level walking take a step sequentially to be to this 4 class conversion behavior, the alternating taken downstairs is changed：Strong side (the level land row of simulation
Walk) the strong side of-simulation Ipsilateral (level walking)-simulation (level walking to upstairs or downstairs or go up a slope or descending is changed)-simulation
Ipsilateral (upstairs, go downstairs, go up a slope, descending) or simulation Ipsilateral-simulation are good for side-simulation Ipsilateral-simulation and are good for side.For upper slope aspect
Level walking conversion, the conversion of lower slope aspect level walking, change to level walking, downstairs changed to level walking upstairs, takes
Alternating takes a step sequentially to be：The strong side of simulation (upstairs, downstairs, go up a slope, descending)-simulation Ipsilateral (upstairs, downstairs, go up a slope, descending)-
The strong side of simulation (upstairs or downstairs or go up a slope or the conversion of lower slope aspect level walking)-simulation Ipsilateral (level walking) or simulation are suffered from
Side-simulation is good for side-simulation Ipsilateral-simulation and is good for side.
Experimenter is recorded and is sorted out to the intention assessment data that inertial sensor produces, and is converted into being counted
The data file directly read according to analysis software.The recorded separately preservation of intention assessment data of each subject, and it is each different
Gait behavior need to put on corresponding label, so that follow-up intention assessment algorithm carries out the study and classification that have supervision.
Data are preserved hereof frame by frame according to the structure of sensor with certain form.Specifically, single sensor
The data format of a given frame is as shown in table 1：
1 single sensor of table gives the data format of a frame
2 sensors (at the strong side thigh of simulation and shank), give the data of a frame, according to
2 form of table arranges：
22 sensor frame data forms of table
The time series data that 2 sensors produce, form are as shown in table 3：
The time series data form that 32 sensors of table produce
2)：Intention assessment data prediction, i.e., carry out the data collected denoising, and abnormal data removes, repairing
Deng.Specifically：
Denoising is carried out to initial data, abnormal data is rejected, missing data is repaired.
As shown in figure 3, using each step touchdown point as reference, long sequence data is cut, after cutting each
Sample is the sensing data of each step.
Windowing process is carried out to the data after cutting, makes all sample data window sizes equal, is characterized extraction and lays
Basis.
In the sensing data of each step after dicing, required data are extracted.For 8 kinds of conversion behaviors, extract
The sample data of conversion step (being transformed into latter mode, that step that the same side less touch with the ground to foot contacts to earth from previous pattern),
For 5 kinds of stable state behaviors, corresponding same foot is extracted since liftoff to the data contacted to earth next time.
3：Intention assessment data characteristics is extracted, and selection can distinguish different classes of motion characteristics attribute, and extract sample number
According to feature, specifically：
For the intention assessment data of 5 class stable state behaviors, the present invention extracts the numerical characteristics of each step data sample, for
8 class conversion behaviors, (upstairs or downstairs or go up a slope or lower slope aspect level land row when the strong side of present invention extraction or Ipsilateral motor pattern are changed
Walk conversion, also or level walking to upstairs or downstairs or go up a slope or descending is changed) numerical characteristics of data sample.In this way, this
The itd is proposed intension recognizing method of invention, can be before motor pattern conversion, it is possible to motion intention is identified, intelligence
The high level controller of artificial limb can change corresponding control parameter with this result, better control over artificial limb.
Extract data characteristics.During human motion, two legs are alternately to step, in a gait cycle, every
Leg all undergoes two stages, is respectively driving phase and recovery phase, referred to as supports phase and shaking peroid.Wherein a leg is in
When the support phase, another leg is in shaking peroid.In the data of well cutting, the present invention have chosen cut in shaking peroid that
Three thigh, shank and ankle sensors of leg, and the average for the data that three sensors collect, side are extracted respectively
The numerical characteristics such as difference, extreme value.In addition it is also possible to using thering is supervision or unsupervised deep learning method to learn automatically and extract meaning
The internal characteristics of figure identification data.
4：Disaggregated model selects and model training, selects suitable disaggregated model to be trained.Specifically：
Selection support vector machines is classified.Due to SVM (support vector machines) can be good at solving it is non-linear and high
The practical problems such as dimension classification.Therefore the SVM using radial basis function as kernel function is used, intention assessment data are extracted
Feature is classified, and combines the optimized parameter that grid data service chooses SVM with K folding cross-validation methods, is prevented over-fitting or is owed
The generation of fitting phenomenon, optimal disaggregated model is obtained with this.The key link of SVM is just to solve for following optimization problem：
s.t y(ω·xi+b)≥1-ξi, i=1,2,3 ..., n
ξ >=0,
Wherein, ξiReferred to as slack variable, C are penalty.Kernel function is one of key factor of SVM, radial direction base core letter
Several forms is as follows：
Wherein, σ is nuclear radius.
In addition, the sorting techniques such as decision tree, neutral net also may be selected.Or the integrated study classification such as selection random forest
Method.Also or the methods of selected depth neutral net；
5)：Intention assessment is completed, successfully identifies motion intention.
One kind of embodiment described above, simply more preferably embodiment of the invention, those skilled in the art
The usual variations and alternatives that member carries out in the range of technical solution of the present invention should all include within the scope of the present invention.
Claims (10)
1. a kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods, it is characterised in that the method passes through
Establish intention assessment database and data are embedded in artificial limb or are bundled in strong side, when the motor pattern conversion for the Ipsilateral for wearing artificial limb
Before, according to embedded artificial limb or be bundled in the time series data that the sensor of strong side produces, to the motion intention of disabled people patient into
Row precognition identification exactly.
2. according to the method described in claim 1, it is characterized in that, the described method includes：
S1：Intention assessment data acquisition simultaneously establishes database:Subject dresses inertial sensor, takes a step according to what is pre-defined
Sequential movements, collect its exercise data and establish database；
S2：Intention assessment data prediction：The data collected are carried out with denoising, and abnormal data removes and repairing；
S3：Intention assessment data characteristics is extracted：Selection can distinguish different classes of motion characteristics attribute, and extract sample data spy
Sign；
S4：Disaggregated model selects and model training, selects suitable disaggregated model to be trained；
S5：Intention assessment is completed, successfully identifies motion intention.
3. according to the method described in claim 2, it is characterized in that, the S1 is specifically included：
S11：2n subjects are selected, male to female ratio is equal, and wherein n is positive integer, and the subject of each health is in simulation Ipsilateral
For thigh with the strong side of simulation with dressing inertial sensor respectively at shank and ankle, the inertial sensor is able to record human body
The acceleration and angular speed kinematics information at each moment during movement, and with the when ordinal number of time cumulation generation multiple data channel
According to data preserve hereof frame by frame according to the structure of sensor；
S12：Alternating according to pre-defining is taken a step sequentially, simulates disabled person's stable state gait behavior and conversion gait row successively
For the stable state gait behavior includes level walking, and upstairs, downstairs, ascents and descents, the conversion gait behavior includes level land
Walk to conversion of going up a slope, level walking to descending conversion, level walking to conversion upstairs, level walking to conversion downstairs, go up a slope
To level walking conversion, the conversion of lower slope aspect level walking, change to level walking and downstairs changed to level walking upstairs；
S13：The intention assessment data produced to inertial sensor are recorded and sorted out, and be converted into can be soft by data analysis
The data file that part is directly read, the recorded separately preservation of intention assessment data of each subject, and each different gait row
To need to put on corresponding label, the study and classification that carry out having supervision for follow-up intention assessment algorithm.
4. according to the method described in claim 3, it is characterized in that, the S2 is specifically included：
S21：Denoising is carried out to initial data, abnormal data is rejected, missing data is repaired；
S22：With each step since liftoff to contacting to earth, long sequence is first cut into one small step of a small step, it is then small one again
Step the inside adding window cuts into the window of 45 frames；
S23：Windowing process is carried out to the data after cutting, makes all sample data window sizes equal, is used in the sample
Data be 45 frame data since less touch with the ground backward, the window of all data is all equal, and what length of window was chosen is
45 frames, length is minimum, and every 45 frame is 1/4 gait cycle, and the time is one section of swing since a less touch with the ground
Time, tiptoe do not contact to earth also, and the windowing process is specially so that all data are all 45 frames since less touch with the ground backward
Data；
S24：Required data are extracted in data in the window after well cutting again..
5. according to the method described in claim 4, it is characterized in that, for the sample of conversion behavior, extraction conversion step in S24
Notebook data, the conversion step is to be transformed into latter mode, the process that the same side less touch with the ground to foot contacts to earth from previous pattern；It is right
In stable state behavior, corresponding same foot is extracted since liftoff to the data contacted to earth next time.
6. according to the method described in claim 5, it is characterized in that, the S3 is specifically included：
S31：For conversion behavior, the data in that leg of swing phase when extraction conversion walks, for stable state behavior, extraction
The identical data of the swing phase of same foot, it is ensured that all data are consistent in sensing station, length of window；
S32：During extracting data characteristics, two legs alternately step, and in a gait cycle, every leg all undergoes two ranks
Section, is respectively driving phase and recovery phase, wherein a leg is in driving phase, another leg is in recovery phase, in S24
In the data of well cutting, three thigh, shank and ankle sensors in shaking peroid are chosen, and extract three sensings respectively
Average, variance and the extreme value for the data that device collects.
7. according to the method described in claim 2, it is characterized in that, the S4 includes：
S41：Selection support vector machines is classified；
S42：The SVM using radial basis function as kernel function is used, the feature extracted to intention assessment data is classified；
S43：The optimized parameter of SVM in S42 is chosen with reference to grid data service and K folding cross-validation methods, over-fitting is prevented or owes to intend
The generation of phenomenon is closed, optimal disaggregated model is obtained with this.
8. the method according to the description of claim 7 is characterized in that SVM method for solving is as follows in S43：
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>|</mo>
<mo>|</mo>
<mi>&omega;</mi>
<mo>|</mo>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>+</mo>
<mi>C</mi>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>&xi;</mi>
<mi>i</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
s.t y(ω·xi+b)≥1-ξi, i=1,2,3 ..., n
ξ >=0,
Wherein, ξiReferred to as slack variable, C are penalty, and the form of Radial basis kernel function is as follows：
Wherein, σ is nuclear radius.
9. according to the method described in claim 3, it is characterized in that, it is right side that strong side is simulated in the S11, simulation Ipsilateral is a left side
Side.
10. method according to claim 6, it is characterised in that, can be to movement in the S31 when motor pattern is changed
Intention is identified, and the high level controller of intelligent artificial limb can change corresponding control parameter in advance with this result, preferably control
False making limb.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711404559.0A CN107918492A (en) | 2017-12-22 | 2017-12-22 | A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711404559.0A CN107918492A (en) | 2017-12-22 | 2017-12-22 | A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107918492A true CN107918492A (en) | 2018-04-17 |
Family
ID=61893932
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711404559.0A Pending CN107918492A (en) | 2017-12-22 | 2017-12-22 | A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107918492A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109009148A (en) * | 2018-08-24 | 2018-12-18 | 广东工业大学 | A kind of gait function appraisal procedure |
| CN111325287A (en) * | 2020-03-17 | 2020-06-23 | 北京理工大学 | Foot touchdown detection method of humanoid robot |
| CN111568615A (en) * | 2020-04-16 | 2020-08-25 | 南方科技大学 | Electric artificial limb system and electric artificial limb control method |
| CN113314209A (en) * | 2021-06-11 | 2021-08-27 | 吉林大学 | Human body intention identification method based on weighted KNN |
| CN113459102A (en) * | 2021-07-09 | 2021-10-01 | 郑州大学 | Human upper limb intention identification method based on projection reconstruction |
| CN113520683A (en) * | 2021-07-08 | 2021-10-22 | 中国科学技术大学 | Lower limb prosthesis control system and method based on simulation learning |
| CN114831627A (en) * | 2022-03-17 | 2022-08-02 | 吉林大学 | Lower limb prosthesis movement identification method based on three decision trees |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104729507A (en) * | 2015-04-13 | 2015-06-24 | 大连理工大学 | Gait recognition method based on inertial sensor |
| CN104983489A (en) * | 2015-07-28 | 2015-10-21 | 河北工业大学 | Road condition identifying method for lower limb prosthesis walking |
| CN106156524A (en) * | 2016-07-29 | 2016-11-23 | 东北大学 | A kind of online gait planning system and method for Intelligent lower limb power assisting device |
| CN106821391A (en) * | 2017-03-23 | 2017-06-13 | 北京精密机电控制设备研究所 | Body gait acquisition analysis system and method based on inertial sensor information fusion |
-
2017
- 2017-12-22 CN CN201711404559.0A patent/CN107918492A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104729507A (en) * | 2015-04-13 | 2015-06-24 | 大连理工大学 | Gait recognition method based on inertial sensor |
| CN104983489A (en) * | 2015-07-28 | 2015-10-21 | 河北工业大学 | Road condition identifying method for lower limb prosthesis walking |
| CN106156524A (en) * | 2016-07-29 | 2016-11-23 | 东北大学 | A kind of online gait planning system and method for Intelligent lower limb power assisting device |
| CN106821391A (en) * | 2017-03-23 | 2017-06-13 | 北京精密机电控制设备研究所 | Body gait acquisition analysis system and method based on inertial sensor information fusion |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109009148A (en) * | 2018-08-24 | 2018-12-18 | 广东工业大学 | A kind of gait function appraisal procedure |
| CN111325287A (en) * | 2020-03-17 | 2020-06-23 | 北京理工大学 | Foot touchdown detection method of humanoid robot |
| CN111568615A (en) * | 2020-04-16 | 2020-08-25 | 南方科技大学 | Electric artificial limb system and electric artificial limb control method |
| CN113314209A (en) * | 2021-06-11 | 2021-08-27 | 吉林大学 | Human body intention identification method based on weighted KNN |
| CN113314209B (en) * | 2021-06-11 | 2023-04-18 | 吉林大学 | Human body intention identification method based on weighted KNN |
| CN113520683A (en) * | 2021-07-08 | 2021-10-22 | 中国科学技术大学 | Lower limb prosthesis control system and method based on simulation learning |
| CN113459102A (en) * | 2021-07-09 | 2021-10-01 | 郑州大学 | Human upper limb intention identification method based on projection reconstruction |
| CN113459102B (en) * | 2021-07-09 | 2022-07-05 | 郑州大学 | Human upper limb intention identification method based on projection reconstruction |
| CN114831627A (en) * | 2022-03-17 | 2022-08-02 | 吉林大学 | Lower limb prosthesis movement identification method based on three decision trees |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107918492A (en) | A kind of human motion in face of Intelligent lower limb artificial limb is intended to precognition recognition methods | |
| CN110537922B (en) | Human body walking process lower limb movement identification method and system based on deep learning | |
| CN108831527B (en) | User motion state detection method and device and wearable device | |
| CN108244744B (en) | Motion state identification method, sole and shoe | |
| CN101807245B (en) | Artificial neural network-based multi-source gait feature extraction and identification method | |
| CN104983489B (en) | Road conditions recognition methods during artificial leg walking | |
| CN110334573B (en) | Human motion state discrimination method based on dense connection convolutional neural network | |
| CN107832686A (en) | Merge the lower limb motion mode recognition methods of surface myoelectric and acceleration signal | |
| CN108683724A (en) | A kind of intelligence children's safety and gait health monitoring system | |
| CN103400123A (en) | Gait type identification method based on three-axis acceleration sensor and neural network | |
| CN105303183B (en) | A kind of child posture discriminance analysis system and method based on wearable device | |
| CN108446733A (en) | A kind of human body behavior monitoring and intelligent identification Method based on multi-sensor data | |
| CN111506189B (en) | Motion mode prediction and switching control method for complex motion of human body | |
| CN113314209B (en) | Human body intention identification method based on weighted KNN | |
| Lee et al. | Optimal time-window derivation for human-activity recognition based on convolutional neural networks of repeated rehabilitation motions | |
| CN107479702A (en) | A kind of human emotion's dominance classifying identification method using EEG signals | |
| CN111611859B (en) | Gait recognition method based on GRU | |
| Peng et al. | Locomotion prediction for lower limb prostheses in complex environments via sEMG and inertial sensors | |
| Zhang et al. | Pathological gait detection of Parkinson's disease using sparse representation | |
| CN114099234A (en) | Intelligent rehabilitation robot data processing method and system for assisting rehabilitation training | |
| Yang et al. | Inertial sensing for lateral walking gait detection and application in lateral resistance exoskeleton | |
| CN108433728A (en) | A method of million accidents of danger are fallen based on smart mobile phone and ANN identification construction personnel | |
| CN108717548A (en) | A kind of increased Activity recognition model update method of facing sensing device dynamic and system | |
| Duong et al. | Ecological validation of machine learning models for spatiotemporal gait analysis in free-living environments using instrumented insoles | |
| CN205031391U (en) | Road condition recognition device of power type artificial limb |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180417 |
|
| RJ01 | Rejection of invention patent application after publication |