CN103006358A - Control method of myoelectric artificial hand - Google Patents

Abstract

The invention relates to a control method of a myoelectric artificial hand. The control method comprises an action recognition method based on adaptive learning of myoelectric signals and a fuzzy control method of grip strength. According to the action recognition method based on adaptive learning of myoelectric signals, opening myoelectric signals EMG1 (electromyography 1) and closing myoelectric signals EMG2 collected from one pair of antagonistic muscles of a human arm are used as input signals after being amplified, rectified and filtered, self-adaption to myoelectric signals of a human body is realized by updating an adaptive scale factor, and the expected grip strength of the artificial hand is output according to the degree of contraction of arm muscles of a user. The fuzzy control method of grip strength is used for realizing tracking of the expected grip strength of the artificial hand and thus realizing control on the grip strength of the artificial hand according to the strength of myoelectric signals of the arm of the user with the artificial hand. Control on the artificial hand can be realized only by adhering two myoelectric sensors to one pair of antagonistic muscles of the arm of the user, turning on a power supply of the artificial hand, and lifting and lowering the wrist of the user.

Description

A kind of control method of EMG-controlling prosthetic hand

One, technical field

The present invention relates to a kind of control method of EMG-controlling prosthetic hand, be used for the control that the hand muscle electricity is done evil through another person, belong to the special instrument control field.

Two, background technology

In EMG-controlling prosthetic hand control, although the recognition decision method of tradition electromyographic signal has certain effectiveness, but do not take into full account different wearers or samely be worn on the problem that electromyographic signal there are differences under the different condition, cause the parameter adjustment of doing evil through another person and when installation is worn, will vary with each individual, have certain limitation in actual the use; At grip controlling party face, some control algolithms have obtained preferably grip tracking control effect based on comparatively perfect sensor-based system is installed on doing evil through another person, yet, actual EMG-controlling prosthetic hand causes present business-like EMG-controlling prosthetic hand to be difficult to realize accurately power control owing to the restriction of size, size, production cost etc. can't be installed sufficient sensor.

Three, summary of the invention

The object of the present invention is to provide a kind ofly have adaptive learning action recognition ability, the control method of the EMG-controlling prosthetic hand of the parameter adjustment that when installation is worn, will vary with each individual of can avoiding doing evil through another person.

The object of the present invention is achieved like this:

Step 1: get with the first myoelectric sensor of the first AD converter with the second myoelectric sensor of the second AD converter, described the first myoelectric sensor and the second myoelectric sensor are respectively applied to gather the closed electromyographic signal EMG2 in the electromyographic signal of opening a pair of Antagonistic muscle on electromyographic signal EMG1 and the arm in the electromyographic signal of a pair of Antagonistic muscle on the arm, make Max _E1, Max _E2Initial value be 0, Min _E1, Min _E2Initial value be 2 ^M, M is the figure place of AD converter, Max _E1, Max _E2Be respectively through the maximum that opens electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing, Min _E1, Min _E2Be respectively through the minima of opening electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing,

Step 2: electromyographic signal adaptive learning action recognition, the expectation grip F that output is done evil through another person _d, detailed step is as follows:

Step 2.1 is utilized the first myoelectric sensor to gather to open electromyographic signal EMG1 in the electromyographic signal of a pair of Antagonistic muscle on the arm, utilize the second myoelectric sensor to gather closed electromyographic signal EMG2 in the electromyographic signal of a pair of Antagonistic muscle on the arm, electromyographic signal EMG1 and closed electromyographic signal EMG2 amplify to opening respectively again, rectification, filtering, obtain pretreated electromyographic signal EMG1 and the pretreated closed electromyographic signal EMG2 of opening

Step 2.2: upgrade respectively the pretreated maximum Max that opens electromyographic signal EMG1 _E1, the pretreated minimum M in that opens electromyographic signal EMG1 _E1, pretreated closed electromyographic signal EMG2 maximum Max _E2And the minimum M in of pretreated closed electromyographic signal EMG2 _E2, update method is as follows:

${\mathrm{Max}}_{E1}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E1},& {\mathrm{Max}}_{E1}\&GreaterEqual;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Max}}_{E1}<\mathrm{EMG}1\end{array}\right.$

${\mathrm{Min}}_{\mathrm{<figure-callout\; id="1"\; label="E"\; filenames="HDA00002660360500011.png,HDA00002660360500014.png"\; state="\{\{state\}\}">E</figure-callout>}1}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E1},& {\mathrm{Min}}_{E1}\&le;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Min}}_{E1}>\mathrm{EMG}1\end{array}\right.$

${\mathrm{Max}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E2},& {\mathrm{Max}}_{E2}\&GreaterEqual;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Max}}_{E2}<\mathrm{EMG}2\end{array}\right.$

${\mathrm{Min}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E2},& {\mathrm{Min}}_{E2}\&le;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Min}}_{E2}>\mathrm{EMG}2\end{array}\right.$

Calculate respectively the self adaptation scale factor K of opening electromyographic signal _E1And the self adaptation scale factor K of closed electromyographic signal _E2:

$\left\{\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="K\; E"\; filenames="HDA00002660360500011.png,HDA00002660360500014.png"\; state="\{\{state\}\}">K</figure-callout>}}_E=\frac{1}{{\mathrm{Max}}_{E1}-{\mathrm{Min}}_{E1}}\\ K_{E2}=\frac{1}{{\mathrm{Max}}_{E2}-{\mathrm{Min}}_{E2}}\end{array}\right.$

Step 2.3: calculate the current stretching degree E that collects electromyographic signal ₁With closed degree E ₂:

E ₁＝(EMG1-Min _E1)×K _E1

E ₂＝(EMG2-Min _E2)×K _E2

Step 2.4: according to the stretching degree E of electromyographic signal ₁With closed degree E ₂Move differentiation, the differentiation result is E, and for opening, for closed, E=0 moves as stopping corresponding to the expectation of doing evil through another person corresponding to the expectation action of doing evil through another person in E＜0 corresponding to the expectation action of doing evil through another person in E＞0,

E＝(E ₁-E ₂)×K _E

Wherein, K _EBe the conversion proportion factor, maximal grip strength design load F equals on the numerical value to do evil through another person _Max, be used for the intensity-conversion of electromyographic signal is the corresponding expectation grip of doing evil through another person,

Step 2.5: the arm of eliminating the wearer that does evil through another person owing to subtle disruption causes the misoperation of doing evil through another person, is exported the expectation grip F that does evil through another person under relaxation state _d,

F _d＝E×f(E)

Wherein $f\left(E\right)=1+\frac{1}{2}[\mathrm{sgn}(E-E_0)-\mathrm{sgn}(E+E_0)],$ $\mathrm{sgn}(E-E_0)=\left\{\begin{array}{cc}1,& E>E_0\\ 0,& E=E_0\\ -1,& E<E_0\end{array}\right.,$ $\mathrm{sgn}(E+E_0)=\left\{\begin{array}{cc}1,& E>-E_0\\ 0,& E=-E_0\\ -1,& E<-E_0\end{array}\right.,$ E ₀=0.05 * K _E, E ₀Be the maximum interference signal value of setting,

Step 3: by the control signal y that the grip fuzzy controller obtains doing evil through another person, detailed step is as follows:

Step 3.1: the current grip signal F that does evil through another person by the 3rd AD converter collection _n, by the do evil through another person current feedback signal I of upper direct current generator of the 4th AD converter collection, calculate grip tracking error Δ F and grip rate of change

ΔF＝F _d-F _n

${\stackrel{\&CenterDot;}{F}}_n=F_n-F_n^{\&prime;}$

F wherein _n' be the last grip signal of doing evil through another person that collects, F during the 1st execution in step 3.1 _n'=0,

Step 3.2: with grip tracking error Δ F and grip rate of change Input grip fuzzy controller, the control signal y that the output of grip fuzzy controller is done evil through another person,

Step 4: as input, the current protection link is output as U to the current protection link with grip fuzzy controller output signal y and the feedback current I of upper motor of doing evil through another person,

U＝y×f(I)

Wherein $f\left(I\right)=\frac{1}{2}[1-\mathrm{sgn}(I-I_m)],$ $\mathrm{sgn}(I-I_m)=\left\{\begin{array}{cc}1,& I>I_m\\ 0,& I=I_m\\ -1,& I<I_m\end{array}\right.,$ I _mBe the motor current at continuous torque of indicating in the motor operation instruction,

Step 5: the output signal U of current protection link is carried out being used for after the power amplification driving direct current generator on doing evil through another person, thus the grip that control is done evil through another person,

Step 6: return step 2.1.

Only two myoelectric sensors need to be fitted on a pair of Antagonistic muscle of wearer's arm during use, open the power supply of doing evil through another person, lift on the wearer and lower wrist can realize control to doing evil through another person.

Compared with prior art, the present invention has following advantage:

1, electromyographic signal adaptive learning action recognition device is by simple electromyographic signal adaptive learning, the amplitude range of real-time update two-way electromyographic signal, stretching degree and the closed degree of electromyographic signal opened in calculating, compare with traditional method that the two-way electromyographic signal is directly subtracted each other, avoided different wearers or same being worn under the different condition to there are differences owing to electromyographic signal, the problem of the parameter adjustment that will vary with each individual when installation is worn causes doing evil through another person, improve the universality that is fit to, also greatly reduced the study adaptive time of wearer to doing evil through another person simultaneously.

2, grip FUZZY ALGORITHMS FOR CONTROL, in the situation that only have force transducer realized doing evil through another person preferably grip control, the algorithm complex of design is low, and amount of calculation is little, need not high-end processor and just can realize the flexible control of doing evil through another person, be suitable for the application of commercialization in doing evil through another person.

3, prosthetic hand control method operating habit with human body natural's hands when practical operation of design is identical, the brief adaptive time of wearer to doing evil through another person.

4, in the action recognition device based on the electromyographic signal adaptive learning of design the debounce link is arranged, reduced the probability of happening of the misoperation of doing evil through another person.

5, current protection link in the grip controller of doing evil through another person of design prevents from damaging because of the long-time stall motor of motor.

Four, description of drawings

Fig. 1 is the composition frame chart of the control method of a kind of EMG-controlling prosthetic hand of the present invention.

Fig. 2 is the action identification method composition frame chart based on the electromyographic signal adaptive learning of the control method of a kind of EMG-controlling prosthetic hand of the present invention.

Fig. 3 is the membership function figure of Δ F of the control method of a kind of EMG-controlling prosthetic hand of the present invention.

Fig. 4 is the control method of a kind of EMG-controlling prosthetic hand of the present invention

Membership function figure.

Five, the specific embodiment

Say in detail implementation step of the present invention below in conjunction with accompanying drawing.

A kind of control method of EMG-controlling prosthetic hand comprises action identification method and grip fuzzy control method based on the electromyographic signal adaptive learning, control flow as shown in Figure 1, concrete steps are as follows:

Step 1: get with the first myoelectric sensor of the first AD converter with the second myoelectric sensor of the second AD converter, described the first myoelectric sensor and the second myoelectric sensor are respectively applied to gather the closed electromyographic signal EMG2 in the electromyographic signal of opening a pair of Antagonistic muscle on electromyographic signal EMG1 and the arm in the electromyographic signal of a pair of Antagonistic muscle on the arm, make Max _E1, Max _E2Initial value be 0, Min _E1, Min _E2Initial value be 2 ^M, M is the figure place of AD converter, Max _E1, Max _E2Be respectively through the maximum that opens electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing, Min _E1, Min _E2Be respectively through the minima of opening electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing,

Step 2: electromyographic signal adaptive learning action recognition, action recognition flow process are exported the expectation grip F that does evil through another person as shown in Figure 2 _d, detailed step is as follows:

Step 2.1 is utilized the first myoelectric sensor to gather to open electromyographic signal EMG1 in the electromyographic signal of a pair of Antagonistic muscle on the arm, utilize the second myoelectric sensor to gather closed electromyographic signal EMG2 in the electromyographic signal of a pair of Antagonistic muscle on the arm, electromyographic signal EMG1 and closed electromyographic signal EMG2 amplify to opening respectively again, rectification, filtering, obtain pretreated electromyographic signal EMG1 and the pretreated closed electromyographic signal EMG2 of opening

Step 2.2: upgrade respectively the pretreated maximum Max that opens electromyographic signal EMG1 _E1, the pretreated minimum M in that opens electromyographic signal EMG1 _E1, pretreated closed electromyographic signal EMG2 maximum Max _E2And the minimum M in of pretreated closed electromyographic signal EMG2 _E2, update method is as follows:

${\mathrm{Max}}_{E1}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E1},& {\mathrm{Max}}_{E1}\&GreaterEqual;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Max}}_{E1}<\mathrm{EMG}1\end{array}\right.$

${\mathrm{Min}}_{\mathrm{<figure-callout\; id="1"\; label="E"\; filenames="HDA00002660360500011.png,HDA00002660360500014.png"\; state="\{\{state\}\}">E</figure-callout>}1}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E1},& {\mathrm{Min}}_{E1}\&le;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Min}}_{E1}>\mathrm{EMG}1\end{array}\right.$

${\mathrm{Max}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E2},& {\mathrm{Max}}_{E2}\&GreaterEqual;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Max}}_{E2}<\mathrm{EMG}2\end{array}\right.$

${\mathrm{Min}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E2},& {\mathrm{Min}}_{E2}\&le;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Min}}_{E2}>\mathrm{EMG}2\end{array}\right.$

Calculate respectively the self adaptation scale factor K of opening electromyographic signal _E1And the self adaptation scale factor K of closed electromyographic signal _E2:

${\mathrm{<figure-callout\; id="1"\; label="K\; E"\; filenames="HDA00002660360500011.png,HDA00002660360500014.png"\; state="\{\{state\}\}">K</figure-callout>}}_E=\frac{1}{{\mathrm{Max}}_{E1}-{\mathrm{<figure-callout\; id="1"\; label="Min\; E"\; filenames="HDA00002660360500011.png,HDA00002660360500014.png"\; state="\{\{state\}\}">Min</figure-callout>}}_E}$

$K_{E2}=\frac{1}{{\mathrm{Max}}_{E2}-{\mathrm{Min}}_{E2}}$

Step 2.3: calculate the current stretching degree E that collects electromyographic signal ₁With closed degree E ₂:

E ₁＝(EMG1-Min _E1)×K _E1

E ₂＝(EMG2-Min _E2)×K _E2

Step 2.4: according to the stretching degree E of electromyographic signal ₁With closed degree E ₂Move differentiation, the differentiation result is E, and for opening, for closed, E=0 moves as stopping corresponding to the expectation of doing evil through another person corresponding to the expectation action of doing evil through another person in E＜0 corresponding to the expectation action of doing evil through another person in E＞0,

E＝(E ₁-E ₂)×K _E

Wherein, K _EBe the conversion proportion factor, maximal grip strength design load F equals on the numerical value to do evil through another person _Max, be used for the intensity-conversion of electromyographic signal is the corresponding expectation grip of doing evil through another person, the maximal grip strength design load of for example ought doing evil through another person F _MaxDuring=20N, K _E=20,

Step 2.5: the arm of eliminating the wearer that does evil through another person owing to subtle disruption causes the misoperation of doing evil through another person, is exported the expectation grip F that does evil through another person under relaxation state _d,

F _d＝E×f(E)

Wherein $f\left(E\right)=1+\frac{1}{2}[\mathrm{sgn}(E-E_0)-\mathrm{sgn}(E+E_0)],$ $\mathrm{sgn}(E-E_0)=\left\{\begin{array}{cc}1,& E>E_0\\ 0,& E=E_0\\ -1,& E<E_0\end{array}\right.,$ $\mathrm{sgn}(E+E_0)=\left\{\begin{array}{cc}1,& E>-E_0\\ 0,& E=-E_0\\ -1,& E<-E_0\end{array}\right.,$ E ₀Be the maximum interference signal value of setting, establishing method is: E ₀=0.05 * K _E

Step 3: by the control signal y that the grip fuzzy controller obtains doing evil through another person, detailed step is as follows:

Step 3.1: the current grip signal F that does evil through another person by the 3rd AD converter collection _n, by the do evil through another person current feedback signal I of upper direct current generator of the 4th AD converter collection, calculate grip tracking error Δ F and grip rate of change

ΔF＝F _d-F _n

${\stackrel{\&CenterDot;}{F}}_n=F_n-F_n^{\&prime;}$

F wherein _n' be the last grip signal of doing evil through another person that collects, F during the 1st execution in step 3.1 _n'=0,

Step 3.2: with grip tracking error Δ F and grip rate of change

Input grip fuzzy controller, the control signal y that the output of grip fuzzy controller is done evil through another person, described grip fuzzy controller adopts fuzzy controller known in this field and commonly used, and is specific as follows:

Step 3.2.1: the degree of membership A that calculates Δ F _i(Δ F), adopt the triangle membership function, the membership function of Δ F as shown in Figure 3, Δ F is divided into 7 grades: negative large (NL), negative in (NM), negative little (NS), zero (ZO), just little (PS), center (PM), honest (PL), f ₀, f ₆Correspond respectively to maximum, the minima of Δ F,

Wherein, f _Δ=(f ₆-f ₀)/6, a _{Δ F}=1/f _Δ,

Step 3.2.2: calculate

Degree of membership

Adopt the triangle membership function,

Membership function as shown in Figure 4, Be divided into 5 grades: negative large (NL), negative little (NS), zero (ZO), just little (PS), honest (PL),

Correspond respectively to

Maximum, minima,

Wherein, ${\stackrel{\&CenterDot;}{f}}_{\mathrm{\&Delta;}}=({\stackrel{\&CenterDot;}{f}}_4-{\stackrel{\&CenterDot;}{f}}_0)/4,$ $a_{{\stackrel{\&CenterDot;}{F}}_n}=1/{\stackrel{\&CenterDot;}{f}}_{\mathrm{\&Delta;}}$

Step 3.2.3: computation rule activity α _Ij,

${\mathrm{\&alpha;}}_{\mathrm{ij}}=\min (1,A_i\left(\mathrm{\&Delta;F}\right)+A_j\left({\stackrel{\&CenterDot;}{F}}_n\right)),$ i＝0...6，j＝0...4

Step 3.2.4: the inquiry fuzzy reasoning table, the value that obtains the ij rule is f _Ij,

Step 3.2.5: adopt the output y of calculated with weighted average method grip fuzzy controller,

$y=\underset{i=0}{\overset{6}{\mathrm{\&Sigma;}}}\underset{j=0}{\overset{4}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_{\mathrm{ij}}\&times;f_{\mathrm{ij}}/\left(\underset{i=0}{\overset{6}{\mathrm{\&Sigma;}}}\underset{j=0}{\overset{4}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_{\mathrm{ij}}\right)$

Step 4: as input, the current protection link is output as U to the current protection link with grip fuzzy controller output signal y and the feedback current I of upper motor of doing evil through another person,

U＝y×f(I)

Wherein $f\left(I\right)=\frac{1}{2}[1-\mathrm{sgn}(I-I_m)],$ $\mathrm{sgn}(I-I_m)=\left\{\begin{array}{cc}1,& I>I_m\\ 0,& I=I_m\\ -1,& I<I_m\end{array}\right.,$ I _mBe the motor current at continuous torque of indicating in the motor operation instruction,

Step 5: the output signal U of current protection link is carried out being used for after the power amplification driving direct current generator on doing evil through another person, thus the grip that control is done evil through another person,

Step 6: return step 2.1.

Claims (1)

1. the control method of an EMG-controlling prosthetic hand, concrete steps are as follows:

Step 1: get with the first myoelectric sensor of the first AD converter with the second myoelectric sensor of the second AD converter, described the first myoelectric sensor and the second myoelectric sensor are respectively applied to gather the closed electromyographic signal EMG2 in the electromyographic signal of opening a pair of Antagonistic muscle on electromyographic signal EMG1 and the arm in the electromyographic signal of a pair of Antagonistic muscle on the arm, make Max _E1, Max _E2Initial value be 0, Min _E1, Min _E2Initial value be 2 ^M, M is the figure place of AD converter, Max _E1, Max _E2Be respectively through the maximum that opens electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing, Min _E1, Min _E2Be respectively through the minima of opening electromyographic signal EMG1 and closed electromyographic signal EMG2 after amplification, rectification, the filter preprocessing,

Step 2: electromyographic signal adaptive learning action recognition, the expectation grip F that output is done evil through another person _d, detailed step is as follows:

Step 2.1 is utilized the first myoelectric sensor to gather to open electromyographic signal EMG1 in the electromyographic signal of a pair of Antagonistic muscle on the arm, utilize the second myoelectric sensor to gather closed electromyographic signal EMG2 in the electromyographic signal of a pair of Antagonistic muscle on the arm, electromyographic signal EMG1 and closed electromyographic signal EMG2 amplify to opening respectively again, rectification, filtering, obtain pretreated electromyographic signal EMG1 and the pretreated closed electromyographic signal EMG2 of opening

Step 2.2: upgrade respectively the pretreated maximum Max that opens electromyographic signal EMG1 _E1, the pretreated minimum M in that opens electromyographic signal EMG1 _E1, pretreated closed electromyographic signal EMG2 maximum Max _E2And the minimum M in of pretreated closed electromyographic signal EMG2 _E2, update method is as follows:

${\mathrm{Max}}_{E1}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E1},& {\mathrm{Max}}_{E1}\&GreaterEqual;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Max}}_{E1}<\mathrm{EMG}1\end{array}\right.$

${\mathrm{Min}}_{E1}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E1},& {\mathrm{Min}}_{E1}\&le;\mathrm{EMG}1\\ \mathrm{EMG}1,& {\mathrm{Min}}_{E1}>\mathrm{EMG}1\end{array}\right.$

${\mathrm{Max}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Max}}_{E2},& {\mathrm{Max}}_{E2}\&GreaterEqual;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Max}}_{E2}<\mathrm{EMG}2\end{array}\right.$

${\mathrm{Min}}_{E2}=\left\{\begin{array}{cc}{\mathrm{Min}}_{E2},& {\mathrm{Min}}_{E2}\&le;\mathrm{EMG}2\\ \mathrm{EMG}2,& {\mathrm{Min}}_{E2}>\mathrm{EMG}2\end{array}\right.$

Calculate respectively the self adaptation scale factor K of opening electromyographic signal _E1And the self adaptation scale factor K of closed electromyographic signal _E2:

$\left\{\begin{array}{c}{K}_{E1}=\frac{1}{{\mathrm{Max}}_{E1}-{\mathrm{Min}}_{E1}}\\ K_{E2}=\frac{1}{{\mathrm{Max}}_{E2}-{\mathrm{Min}}_{E2}}\end{array}\right.$

Step 2.3: calculate the current stretching degree E that collects electromyographic signal ₁With closed degree E ₂:

E ₁＝(EMG1-Min _E1)×K _E1

E ₂＝(EMG2-Min _E2)×K _E2

Step 2.4: according to the stretching degree E of electromyographic signal ₁With closed degree E ₂Move differentiation, the differentiation result is E, and for opening, for closed, E=0 moves as stopping corresponding to the expectation of doing evil through another person corresponding to the expectation action of doing evil through another person in E＜0 corresponding to the expectation action of doing evil through another person in E＞0,

E＝(E ₁-E ₂)×K _E

Wherein, K _EBe the conversion proportion factor, maximal grip strength design load F equals on the numerical value to do evil through another person _Max, be used for the intensity-conversion of electromyographic signal is the corresponding expectation grip of doing evil through another person,

Step 2.5: the arm of eliminating the wearer that does evil through another person owing to subtle disruption causes the misoperation of doing evil through another person, is exported the expectation grip F that does evil through another person under relaxation state _d,

F _d＝E×f(E)

Wherein $f\left(E\right)=1+\frac{1}{2}[\mathrm{sgn}(E-E_0)-\mathrm{sgn}(E+E_0)],$ $\mathrm{sgn}(E-E_0)=\left\{\begin{array}{cc}1,& E>E_0\\ 0,& E=E_0\\ -1,& E<E_0\end{array}\right.,$ $\mathrm{sgn}(E+E_0)=\left\{\begin{array}{cc}1,& E>-E_0\\ 0,& E=-E_0\\ -1,& E<-E_0\end{array}\right.,$ E ₀=0.05 * K _E, E ₀Be the maximum interference signal value of setting,

Step 3: by the control signal y that the grip fuzzy controller obtains doing evil through another person, detailed step is as follows:

Step 3.1: the current grip signal F that does evil through another person by the 3rd AD converter collection _n, by the do evil through another person current feedback signal I of upper direct current generator of the 4th AD converter collection, calculate grip tracking error Δ F and grip rate of change

ΔF＝F _d-F _n

${\stackrel{\&CenterDot;}{F}}_n=F_n-F_n^{\&prime;}$

F wherein _n' be the last grip signal of doing evil through another person that collects, F during the 1st execution in step 3.1 _n'=0,

Step 3.2: with grip tracking error Δ F and grip rate of change

Input grip fuzzy controller, the control signal y that the output of grip fuzzy controller is done evil through another person,

Step 4: as input, the current protection link is output as U to the current protection link with grip fuzzy controller output signal y and the feedback current I of upper motor of doing evil through another person,

U＝y×f(I)

Wherein $f\left(I\right)=\frac{1}{2}[1-\mathrm{sgn}(I-I_m)],$ $\mathrm{sgn}(I-I_m)=\left\{\begin{array}{cc}1,& I>I_m\\ 0,& I=I_m\\ -1,& I<I_m\end{array}\right.,$ I _mBe the motor current at continuous torque of indicating in the motor operation instruction,

Step 5: the output signal U of current protection link is carried out being used for after the power amplification driving direct current generator on doing evil through another person, thus the grip that control is done evil through another person,

Step 6: return step 2.1.

Images