CN109227543A - A kind of limited pattern-based intelligent control method of flexible joint robot in position - Google Patents
A kind of limited pattern-based intelligent control method of flexible joint robot in position Download PDFInfo
- Publication number
- CN109227543A CN109227543A CN201811219207.2A CN201811219207A CN109227543A CN 109227543 A CN109227543 A CN 109227543A CN 201811219207 A CN201811219207 A CN 201811219207A CN 109227543 A CN109227543 A CN 109227543A
- Authority
- CN
- China
- Prior art keywords
- dynamic
- mode
- flexible joint
- pattern
- joint robot
- 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
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
Abstract
The invention discloses a kind of positions to be limited the pattern-based intelligent control method of flexible joint robot, comprising the following steps: the kinetic model and several expectations of establishing flexible joint robot return the universal model of track;Based on actual condition, the limited constraint condition of given position designs corresponding position transfer function, according to the determining constant value nerve network controller group that meets transient performance requirement of the theories of learning design based on reference model;Establish the dynamic mode library that expectation returns locus model;Design pattern-recognition scheme and controller switchover policy based on dynamic mode.This method makes flexible joint robot obtain and utilize Heuristics from complicated task in the case where output variable meets the constraint condition specified, system is realized to the real-time monitoring of dynamic mode and autonomous quickly identification, it ensure that the smooth continuity for controlling input signal in mode handover procedure simultaneously, provide guarantee for the stability of control system.
Description
Technical field
The present invention relates to the flexible joint robot control fields based on pattern-recognition, and in particular to a kind of position is limited soft
The property pattern-based intelligent control method of articulated robot.
Background technique
Mankind's science and technology is maked rapid progress, and the robot of mankind's burden and work can effectively be mitigated therewith by being born, and it is absurd fantastic
Raw and development then further promotes being constantly progressive for human society.Mechanical arm is an important branch of robot, wherein soft
Property articulated robot with its body small, light, low consumption, it is efficient and flexible and convenient the advantages that, become the present and following people and deeply grind
The key areas studied carefully.Research to flexible joint robot includes the research of its kinetic model and grinding for control of which strategy
Study carefully.Especially space flight and aviation, precision machinery and its in terms of, flexible joint robot all occupies very important
Status.And in some special occasions, it can also replace manpower to complete large labor intensity, security risk height and operating environment complexity
Work.Under the overall background of ever-expanding application demand and intelligence manufacture industry, people, which control flexible joint robot, is
The intelligence degree requirement of system is higher and higher, such as, it may require that it can obtain Heuristics in complicated task,
And working efficiency and work quality are improved using the Heuristics of storage, simultaneously because the restriction of actual conditions, state variable
Certain constraint condition must be met.
In recent years, the research of pattern-recognition has obtained the extensive concern of researcher, and the identification of dynamic mode is as mould
Formula identifies most difficult one of the task in field, determines that the theories of learning then provide effective means to solve this respect problem, so
And with the scheme control of the artificial control object of flexible joint machine study application and it is few.Flexible joint machine based on dynamic mode
Device people's control system requires it while the task of execution, being capable of mode performed by real-time monitoring.Meanwhile for real feelings
The considerations of condition, tracking performance flexible joint robot up to standard do not require nothing more than its effectively tracking desired trajectory, to consider yet
The transient response of control system output and steady-state response meet in specified boundary.When making to switch among different tasks,
The switching of controller input signal can have jump, this can damage the driving mechanism of flexible joint robot, thus broken
The stability of bad control system.
Summary of the invention
Flexible joint robot is limited in view of the deficiencies of the prior art, it is an object of the present invention to provide a kind of position to be based on
The intelligent control method of mode proposes dynamic pattern recognition scheme, realizes for the pattern recognition problem under dynamic environment
Mechanical arm system provides the real-time monitoring and quickly identification, the control performance for guarantee joint robot system of dynamic mode
Precondition;Simultaneously on this basis, no limitation problem is converted to by limitation problem is exported using position transfer function, to conversion
Variable afterwards ensure that its stability using liapunov's method, constructs constant value nerve network controller group, devises base
In the controller switchover policy of dynamic mode, successfully solves existing control input jump problem when controller switching, ensure
The safety and stability of control system.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of limited pattern-based intelligent control method of flexible joint robot in position, the method includes following steps
It is rapid:
Step 1, the kinetic model for establishing flexible joint robot and several expectations return locus model;
Step 2, the constraint condition of given flexible joint robot links joint angular displacement:
Wherein, n indicates joint number, q1iThe Angular displacementvector for indicating each link joint, by q11,q12,…,q1nWrite as vector
Mode, q1=(q11,q12,...,q1n)T, q1System output variables before indicating conversion ,-k iWithRespectively indicate i-th of joint angle
The boundary up and down that displacement vector is constrained;
Design position transfer function:
Wherein,By position transfer function above, by limited location variable q1i
Switch to unrestricted location variable s1i:
By s11,s12,…,s1nWrite as vector pattern, s1=(s11,s12,...,s1n)T, s1System output after indicating conversion
Variable;
Therefore, system output dynamic translation relationship is as follows:
Wherein
Expectation is returned track also to be converted accordingly:
By yd1,yd2,…,ydnWrite as vector pattern, yd=(yd1,yd2,...,ydn)T, ydExpectation joint before indicating conversion
Angular displacement;By sd1,sd2,…,sdnWrite as vector pattern, sd=(sd1,sd2,...,sdn)T, sdIndicate the expectation after conversion
Joint angular displacement variable;
Step 3, in conjunction with dynamic surface method, design neural network control device:
Wherein, αi(i=1,2,3) is flexible joint robot system Virtual Controller, and u is the control of whole system
Device, ciIt (i=1,2,3,4) is then respectively αi(i=1,2,3) and in u the normal number control gain matrix designed;Wherein nerve net
NetworkApproach α2In unknown dynamic,It is the estimated value of ideal neural network weight, S2(ψ2) then for vector ψ2
For the Gaussian radial basis function of input;And neural networkFor approaching the unknown dynamic in u,Indicate the mind
Estimated value through network ideal weight, S4(ψ4) then for vector ψ4For the Gaussian radial basis function of input;And zi(i=1,2,
It 3,4) is the median error variable for designing controller process;
Step 4, according to determine the theories of learning, design constant value nerve network controller group:
Wherein,
Wherein, N is the total number of dynamic mode;It is that kth mode can approach after training
α2In unknown dynamic neural network weight constant, t ∈ [t1,t2] it is trainingStable convergence after transient process when network
Period,For radial basis functionInput vector;Similarly,It is training neural network in uObtained weight scalar matrix, t ∈ [t3,t4] it is trainingNet
When network after transient process stable convergence period,For radial basis functionInput
Vector;
Step 5 returns locus model for different expectations, establishes dynamic mode library;
Step 6, design dynamic estimator, decision norm and pattern recognition strategy;
Step 7, controller switchover policy of the design based on dynamic mode:
Wherein,WithIt is illustrated respectively under dynamic mode σThe Virtual Controller α at moment2With controller u,And uσ′It (t) is then in identification processIn Virtual Controller α2With controller u,And uυ(t) right respectively
Under the dynamic mode υ of Ying YuxinVirtual Controller α2With controller u, attenuation coefficient λ > 0 in exponential term;And know
The control of other process inputsAnd uσ′(t) as follows:
Wherein,Indicate trackLearning region,Indicate trackLearning region;
It isWhen the transition gain that is designed to be restored to the track of mode;
And the control input at new dynamic mode υAnd uυ(t) as follows:
Wherein,WithIt is all to control gain matrix accordingly at new model υ.
Further, in step 1, the kinetic model of the flexible joint robot are as follows:
Wherein, q1∈RnFor the Angular displacementvector of link joint, q2∈RnFor the Angular displacementvector of motor, M (q1)∈Rn×n
For the inertial matrix of connecting rod, J ∈ Rn×nFor the motor inertia moment battle array of diagonal positive definite,For coriolis force and centrifugal force
Matrix, g (q1)∈RnFor gravity item, K ∈ Rn×nIt is the stiffness matrix about flexible joint, u ∈ RnIt is inputted for the control of system, q1
For system output.
Further, in step 1, the expectation of every kind of reference model returns track are as follows:
Wherein xd=[xd1 T,xd2 T]TFor the state vector under the mode, yd∈RnFor with reference to output signal, fd(xd) it is
The smooth nonlinear function known.
Further, in step 3, the median error variable z of the design controller processi(i=1,2,3,4) are as follows:
z1=s1-sd
z3=q2-α2f
Wherein, αif(i=1,2,3) it is the filtering virtual controlling variable obtained by following firstorder filter equation respectively:
Wherein, τiIt (i=1,2,3) is the corresponding filter factor in these three firstorder filters respectively.
Further, in step 5, the dynamic mode library:
Wherein,The weight to obtain is trained by RBF neural for the corresponding desired trajectory of kth reference model
Constant, N are the total number of dynamic mode.
Further, in step 6, the dynamic estimator are as follows:
Wherein,For the state of the dynamic estimator under kth mode, xd2For the shape for returning locus model for being tested mode
State variable,It is the neural network weight vector that dynamic mode k is corresponded in dynamic mode library,Be withFor
The radial basis function of input vector, b > 0 are the gain constant designed in dynamic estimator.
Further, in step 6, the decision norm are as follows:
Wherein, residual errorT > 0 is the time constant of design.
Further, in step 6, the pattern recognition strategy are as follows:
Assuming that current dynamic mode is σ, if the norm value of the mode within a certain period of timeIt is set greater than the mode
Fixed threshold epsilonσ, then it can be seen that dynamic mode has changed;It, then can will be following after dynamic mode changes
A period of time in the corresponding dynamic mode υ of Norm minimum value as switching after mode.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1, the present invention successfully solves constrained system and is converted to no constrained system by design position transfer function, so that
After not considering that the mechanical arm intelligent independent control method based on dynamic mode of limitation problem can apply to variable conversion originally
Control system.
2, the present invention is by designing a kind of dynamic pattern recognition scheme, successfully solves control system in a dynamic environment
Pattern classification and identification problem, enable flexible joint robot system to make accurate knowledge when monitored mode changes
Not.
3, the present invention improves controller switching by controller switchover policy of the design based on dynamic mode better
When existing control input jump problem, ensured the stability of control system, improved in system controller handoff procedure
Transient performance.
4, intelligent control side of the flexible joint robot based on reference model that the present invention by design there is position to be limited
Method enables flexible joint robot to obtain from complicated task and utilize Heuristics, to realize more
Intelligent independent control under kind task.
Detailed description of the invention
Fig. 1 is that position of the embodiment of the present invention is limited the pattern-based intelligent control method flow chart of flexible joint robot.
Fig. 2 is that position is limited flexible joint robot system schematic diagram.
Fig. 3 is the phase plane change curve that expectation of the embodiment of the present invention returns track.
Fig. 4 is position transfer function figure.
Fig. 5 is decision of embodiment of the present invention norm change curve.
Fig. 6 is the track following error change curve graph of link joint of the embodiment of the present invention 1.
Fig. 7 is the track following error change curve graph of link joint of the embodiment of the present invention 2.
Fig. 8 is the limited angular displacement trace plot of link joint of the embodiment of the present invention 1.
Fig. 9 is the limited angular displacement trace plot of link joint of the embodiment of the present invention 2.
Figure 10 is that the virtual controlling of link joint of the embodiment of the present invention 1 inputs α21Change curve.
Figure 11 is that the virtual controlling of link joint of the embodiment of the present invention 2 inputs α22Change curve.
Figure 12 is that the control of link joint of the embodiment of the present invention 1 inputs u1Change curve.
Figure 13 is that the control of link joint of the embodiment of the present invention 2 inputs u2Change curve.
Specific embodiment
Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited
In this.
Embodiment:
Present embodiments provide a kind of limited pattern-based intelligent control method of flexible joint robot in position, bulk flow
Journey is as shown in Figure 1, be described in detail the method by taking double link flexible joint robot as an example, detailed implementation process packet
It includes:
Step 1, the kinetic model for establishing double link flexible joint robot:
Wherein q1=[q11,q12] indicate connecting rod rotational angle vector, q2=[q21,q22] indicate motor rotational angle
Vector, M (q1)∈R2×2Indicate the inertial matrix of connecting rod, J ∈ R2×2For the motor inertia moment battle array of diagonal positive definite, in addition,For coriolis force and centrifugal force matrix, g (q1) indicate gravity item.K∈R2×2It is the rigidity about flexible joint
Matrix, while being also the constant matrices of diagonal positive definite, therefore K-1It is also positive definite matrix.U=[u1,u2] indicate that the control of system is defeated
Enter, q1For system output.
The relevant parameter of the double link flexible joint robot model chosen in the present embodiment is respectively as follows:
Wherein, the quality m of connecting rod 1 and connecting rod 21=0.8, m2=2.3, the long l of bar1=l2=1, and g takes 9.8m/s2, j1=
j2=0.5, k1=k2=15.
The universal model of two kinds of reference locus is given in the present embodiment are as follows:
Wherein,Track is returned for the expectation of flexible joint robot links joint angular displacement,Track is returned for the expectation of its link joint angular speed, gives the reference locus under two kinds of dynamic modes,
Then the system dynamic that both dynamic modes can be defined respectively is as follows:
Step 2, the constraint condition of given flexible joint robot links joint angular displacement:
Q is exported according to given flexible joint robot system1And reference locus yd, design output q1Below meeting about
Beam condition:
Wherein n is joint number, q1iEach link joint Angular displacementvector is indicated, by q11,q12,…,q1nWrite as vector pattern,
q1=(q11,q12,...,q1n)T,-k iWithRespectively indicate the boundary up and down that i-th of joint is constrained.
Design position transfer function:
Wherein,By position transfer function above, limited location variable is turned
For unrestricted location variable:
By s11,s12,…,s1nWrite as vector pattern, s1=(s11,s12,...,s1n)T, s1System output after indicating conversion
Variable.
Therefore, system output dynamic translation relationship is as follows:
Wherein
Expectation is returned track also to be converted accordingly:
By yd1,yd2,…,ydnWrite as vector pattern, yd=(yd1,yd2,...,ydn)T, ydExpectation joint before indicating conversion
Angular displacement.By sd1,sd2,…,sdnWrite as vector pattern, sd=(sd1,sd2,...,sdn)T, sdExpectation joint after indicating conversion
Angular displacement variable.
Step 3, in conjunction with dynamic surface method, design neural network control device:
Wherein, αi(i=1,2,3) is flexible joint robot system Virtual Controller, and u is the control of whole system
Device, ciIt (i=1,2,3,4) is then respectively αi(i=1,2,3) and in u the normal number control gain matrix designed;Wherein nerve net
NetworkApproach α2In unknown dynamic,It is the estimated value of ideal neural network weight, S2(ψ2) then for vector ψ2
For the Gaussian radial basis function of input;And neural networkFor approaching the unknown dynamic in u,Indicate the mind
Estimated value through network ideal weight, S4(ψ4) then for vector ψ4For the Gaussian radial basis function of input;And zi(i=1,2,
It 3,4) is the median error variable for designing controller process;What is chosen in example is doubly-linked rod model, so α2By first company
The Virtual Controller α of bar joint21With the Virtual Controller α of second link joint22Composition, i.e. α2=[α21,α22]。
Step 4, according to determine the theories of learning, design constant value nerve network controller group:
Wherein,
Wherein,It is that k mode can approach α after training2In unknown dynamic neural network
Weight constant, t ∈ [t1,t2] it is trainingWhen network after transient process stable convergence period,For radial basis functionInput vector;Similarly,It is to be instructed in u
Practice neural networkObtained weight scalar matrix, t ∈ [t3,t4] it is trainingStablize after transient process when network and receives
The period held back,For radial basis functionInput vector;
Step 5 returns locus model for different expectations, establishes dynamic mode library;
Wherein,For the neural network weight constant of dynamic mode k.
Step 6, design dynamic estimator:
Wherein, subscript k indicates pattern class corresponding to dynamic estimator.For the state of dynamic estimator, xd1And xd2
It is then the system mode of tested reference locus, b > 0 is the gain constant of design.
Then design decision norm:
Wherein, residual errorT > 0 is the time constant of design.
Last design pattern recognition strategy:
Assuming that current dynamic mode is σ, if the norm value of the mode within a certain period of timeIt is set greater than the mode
Fixed threshold epsilonσ, then it is considered that dynamic mode has changed;After dynamic mode changes, we can will be connect down
The corresponding dynamic mode υ of Norm minimum value is as the mode after switching in a period of time come.
Step 8, controller switchover policy of the design based on dynamic mode:
Wherein,WithIt is illustrated respectively under dynamic mode σThe Virtual Controller α at moment2With controller u,And uσ′It (t) is then in identification processIn Virtual Controller α2With controller u,And uυ(t) respectively
Corresponding under new dynamic mode υVirtual Controller α2With controller u, attenuation coefficient λ > 0 in exponential term.And
The control of identification process inputsAnd uσ′It is (t) as follows,
Wherein,Indicate trackLearning region,Indicate trackLearning region;
It isWhen the transition gain that is designed to be restored to the track of mode.
And the control input at new dynamic mode υAnd uυIt is (t) as follows,
Wherein,WithIt is all to control gain matrix accordingly at new model υ.
In the present embodiment, q1、q2、Initial value be q1,1(0)=0.2, q1,2(0)=0,
q2,1(0)=q2,2(0)=0,First neural network of the 1st reference modelIn
Heart point is evenly distributed on [- 1.8,1.8] × [- 1.8,1.8] × [- 2.4,2.4] × [- 2.4,2.4] × [- 3,3] × [- 3,3]
On, width width2_p1=[1.5,1.5,2,2,2.5,2.5]T, number of nodes 4096, and second neural networkCentral point be evenly distributed on [- 1.8,1.8] × [- 1.8,1.8] × [- 3,3] × [- 3,3] × [- 40,44]
On × [- 30,30], width width4_p1=[1.5,1.5,2.5,2.5,35,30]T, number of nodes 4096;2nd reference
First neural network of modeCentral point be evenly distributed on [- 3,3] × [- 3,3] × [- 3,3] × [- 3,3] ×
On [- 6,6] × [- 6,6], width width2_p2=[2.5,2.5,2.5,2.5,5,5]T, number of nodes 4096, and second
Neural networkCentral point be evenly distributed on [- 3,3] × [- 3,3] × [- 4.2,4.2] × [- 4.2,4.2] ×
On [- 60,60] × [- 42,42], width width4_p2=[2.5,2.5,3.5,3.5,50,35]T, number of nodes 4096.
The present embodiment chooses expectation recurrence track and is first switched to the 2nd from the 1st reference model in system operation time section
A reference model, and given output constraint condition isFig. 2 is that position is limited flexible joint robot
System schematic, the two kinds of dynamic modes expectation trained return the phase plane change curve of track as shown in figure 3, Fig. 4 is position
Transfer function figure is set, Fig. 5 is decision of embodiment of the present invention norm change curve, observes Fig. 5, and discovery system existsPlace
Detect that the mode of expectation recurrence track changes,Identify the mode after conversion in place;Fig. 6 and Fig. 7 difference
For the track following error change curve graph of link joint 1 and joint 2, Fig. 8 and Fig. 9 are respectively the angle of link joint 1 and joint 2
Position limited curves figure, this demonstrate designed schemes to solve the problems, such as that output is limited very well;Figure 10 and Figure 11 are respectively
The virtual controlling in link joint 1 and joint 2 inputs α2Change curve, and Figure 12 and Figure 13 be respectively link joint 1 and close
The change curve of the control input u of section 2, by Figure 10 to Figure 13 it is found that being switched according to dynamic pattern recognition scheme and controller
Strategy, system is in the smooth continuity for carrying out ensure that control signal when controller switching, to ensure the stability of system.
The above, only the invention patent preferred embodiment, but the scope of protection of the patent of the present invention is not limited to
This, anyone skilled in the art is in the range disclosed in the invention patent, according to the present invention the skill of patent
Art scheme and its patent of invention design are subject to equivalent substitution or change, belong to the scope of protection of the patent of the present invention.
Claims (8)
1. a kind of position is limited the pattern-based intelligent control method of flexible joint robot, which is characterized in that the method packet
Include following steps:
Step 1, the kinetic model for establishing flexible joint robot and several expectations return locus model;
Step 2, the constraint condition of given flexible joint robot links joint angular displacement:
Wherein, n indicates joint number, q1iThe Angular displacementvector for indicating each link joint, by q11,q12,…,q1nWrite as vector pattern,
q1=(q11,q12,...,q1n)T, q1System output variables before indicating conversion ,-k iWithRespectively indicate i-th of joint angular displacement
The boundary up and down that vector is constrained;
Design position transfer function:
Wherein,By position transfer function above, by limited location variable q1iSwitch to
Unrestricted location variable s1i:
By s11,s12,…,s1nWrite as vector pattern, s1=(s11,s12,...,s1n)T, s1System after indicating conversion, which exports, to be become
Amount;
Therefore, system output dynamic translation relationship is as follows:
Wherein
Expectation is returned track also to be converted accordingly:
By yd1,yd2,…,ydnWrite as vector pattern, yd=(yd1,yd2,...,ydn)T, ydExpectation joint angle position before indicating conversion
It moves;By sd1,sd2,…,sdnWrite as vector pattern, sd=(sd1,sd2,...,sdn)T, sdIndicate the expectation joint after conversion
Angular displacement variable;
Step 3, in conjunction with dynamic surface method, design neural network control device:
Wherein, αi(i=1,2,3) is flexible joint robot system Virtual Controller, and u is the controller of whole system, ci(i
It=1,2,3,4) is then respectively αi(i=1,2,3) and in u the normal number control gain matrix designed;Wherein neural networkApproach α2In unknown dynamic,It is the estimated value of ideal neural network weight, S2(ψ2) then for vector ψ2For
The Gaussian radial basis function of input;And neural networkFor approaching the unknown dynamic in u,Indicate the nerve
The estimated value of network ideal weight, S4(ψ4) then for vector ψ4For the Gaussian radial basis function of input;And zi(i=1,2,3,
It 4) is the median error variable for designing controller process;
Step 4, according to determine the theories of learning, design constant value nerve network controller group:
Wherein,
Wherein, N is the total number of dynamic mode;It is that kth mode can approach α after training2In not
Know dynamic neural network weight constant, t ∈ [t1,t2] it is trainingWhen network after transient process stable convergence time
Section,For radial basis functionInput vector;Similarly,It is u
Middle trained neural networkObtained weight scalar matrix, t ∈ [t3,t4] it is trainingIt is steady after transient process when network
The fixed convergent period,For radial basis functionInput vector;
Step 5 returns locus model for different expectations, establishes dynamic mode library;
Step 6, design dynamic estimator, decision norm and pattern recognition strategy;
Step 7, controller switchover policy of the design based on dynamic mode:
Wherein,WithIt is illustrated respectively under dynamic mode σThe Virtual Controller α at moment2With controller u,And uσ′It (t) is then in identification processIn Virtual Controller α2With controller u,And uυ(t) right respectively
Under the dynamic mode υ of Ying YuxinVirtual Controller α2With controller u, attenuation coefficient λ > 0 in exponential term;And know
The control of other process inputsAnd uσ′(t) as follows:
Wherein,Indicate trackLearning region,Indicate trackLearning region;It isWhen the transition gain that is designed to be restored to the track of mode
And the control input at new dynamic mode υAnd uυ(t) as follows:
Wherein,WithIt is all to control gain matrix accordingly at new model υ.
2. a kind of position according to claim 1 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 1, the kinetic model of the flexible joint robot are as follows:
Wherein, q1∈RnFor the Angular displacementvector of link joint, q2∈RnFor the Angular displacementvector of motor, M (q1)∈Rn×nFor connecting rod
Inertial matrix, J ∈ Rn×nFor the motor inertia moment battle array of diagonal positive definite,For coriolis force and centrifugal force matrix, g
(q1)∈RnFor gravity item, K ∈ Rn×nIt is the stiffness matrix about flexible joint, u ∈ RnIt is inputted for the control of system, q1For system
Output.
3. a kind of position according to claim 1 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 1, the expectation of every kind of reference model returns track are as follows:
Wherein xd=[xd1 T,xd2 T]TFor the state vector under the mode, yd∈RnFor with reference to output signal, fd(xd) it is known
Smooth nonlinear function.
4. a kind of position according to claim 1 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 3, the median error variable z of the design controller processi(i=1,2,3,4) are as follows:
z1=s1-sd
z3=q2-α2f
Wherein, αif(i=1,2,3) it is the filtering virtual controlling variable obtained by following firstorder filter equation respectively:
Wherein, τiIt (i=1,2,3) is the corresponding filter factor in these three firstorder filters respectively.
5. a kind of position according to claim 1 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 5, the dynamic mode library are as follows:
Wherein,The weight constant to obtain, N are trained by RBF neural for the corresponding desired trajectory of kth reference model
For the total number of dynamic mode.
6. a kind of position according to claim 1 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 6, the dynamic estimator are as follows:
Wherein,For the state of the dynamic estimator under kth mode, xd2Become to be tested the state of the recurrence locus model of mode
Amount,It is the neural network weight vector that dynamic mode k is corresponded in dynamic mode library,Be withFor input
The radial basis function of vector, b > 0 are the gain constant designed in dynamic estimator.
7. a kind of position according to claim 6 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 6, the decision norm are as follows:
Wherein, residual errorT > 0 is the time constant of design.
8. a kind of position according to claim 7 is limited the pattern-based intelligent control method of flexible joint robot,
It is characterized in that, in step 6, the pattern recognition strategy are as follows:
Assuming that current dynamic mode is σ, if the norm value of the mode within a certain period of timeGreater than mode setting
Threshold epsilonσ, then it can be seen that dynamic mode has changed;It, then can be by next one after dynamic mode changes
The corresponding dynamic mode υ of Norm minimum value is as the mode after switching in the section time.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811219207.2A CN109227543B (en) | 2018-10-19 | 2018-10-19 | Mode-based intelligent control method for position-limited flexible joint robot |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811219207.2A CN109227543B (en) | 2018-10-19 | 2018-10-19 | Mode-based intelligent control method for position-limited flexible joint robot |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109227543A true CN109227543A (en) | 2019-01-18 |
CN109227543B CN109227543B (en) | 2020-07-28 |
Family
ID=65052430
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811219207.2A Active CN109227543B (en) | 2018-10-19 | 2018-10-19 | Mode-based intelligent control method for position-limited flexible joint robot |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109227543B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109782601A (en) * | 2019-01-31 | 2019-05-21 | 浙江大学 | A kind of coordination mechanical arm adaptive neural network synchronization robust Controller Design method |
CN110427536A (en) * | 2019-08-12 | 2019-11-08 | 深圳忆海原识科技有限公司 | One type brain decision and kinetic control system |
CN112051734A (en) * | 2020-08-18 | 2020-12-08 | 华南理工大学 | Wheeled mobile robot event triggering tracking control method based on deterministic learning |
CN112051733A (en) * | 2020-08-14 | 2020-12-08 | 华南理工大学 | Rigid mechanical arm composite learning control method based on image classification |
CN112947093A (en) * | 2021-04-07 | 2021-06-11 | 长春工业大学 | Robot dispersion robust tracking control method, system and storage medium |
CN113290563A (en) * | 2021-05-31 | 2021-08-24 | 哈尔滨工业大学 | Space target capturing method of variable-rigidity rope-driven snakelike mechanical arm |
CN113433825A (en) * | 2021-06-22 | 2021-09-24 | 广州大学 | Self-adaptive fault-tolerant control method and system of single-link mechanical arm and storage medium |
CN117539153A (en) * | 2023-11-21 | 2024-02-09 | 山东大学 | Upper limb rehabilitation robot self-adaptive control method and system based on definite learning |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1106310A1 (en) * | 1999-11-30 | 2001-06-13 | STMicroelectronics S.r.l. | Electronic circuit for controlling a movement by a Fuzzy cellular architecture |
CN101975955A (en) * | 2010-09-29 | 2011-02-16 | 北京航空航天大学 | Method for generating universal three-dimensional carrier motion trail in GNSS simulator |
CN106078741A (en) * | 2016-06-21 | 2016-11-09 | 华南理工大学 | Based on a determination that the limited performance flexible mechanical arm control method of theory of learning |
CN107160398A (en) * | 2017-06-16 | 2017-09-15 | 华南理工大学 | The safe and reliable control method of Rigid Robot Manipulator is limited based on the total state for determining study |
CN107498554A (en) * | 2017-07-31 | 2017-12-22 | 华南理工大学 | A kind of mechanical arm intelligent independent control method based on dynamic mode |
CN108388123A (en) * | 2018-03-09 | 2018-08-10 | 迪比(重庆)智能科技研究院有限公司 | Self-adaptation control method of multiple anthropomorphic robots in the case where speed can not be surveyed |
-
2018
- 2018-10-19 CN CN201811219207.2A patent/CN109227543B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1106310A1 (en) * | 1999-11-30 | 2001-06-13 | STMicroelectronics S.r.l. | Electronic circuit for controlling a movement by a Fuzzy cellular architecture |
CN101975955A (en) * | 2010-09-29 | 2011-02-16 | 北京航空航天大学 | Method for generating universal three-dimensional carrier motion trail in GNSS simulator |
CN106078741A (en) * | 2016-06-21 | 2016-11-09 | 华南理工大学 | Based on a determination that the limited performance flexible mechanical arm control method of theory of learning |
CN107160398A (en) * | 2017-06-16 | 2017-09-15 | 华南理工大学 | The safe and reliable control method of Rigid Robot Manipulator is limited based on the total state for determining study |
CN107498554A (en) * | 2017-07-31 | 2017-12-22 | 华南理工大学 | A kind of mechanical arm intelligent independent control method based on dynamic mode |
CN108388123A (en) * | 2018-03-09 | 2018-08-10 | 迪比(重庆)智能科技研究院有限公司 | Self-adaptation control method of multiple anthropomorphic robots in the case where speed can not be surveyed |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109782601A (en) * | 2019-01-31 | 2019-05-21 | 浙江大学 | A kind of coordination mechanical arm adaptive neural network synchronization robust Controller Design method |
CN109782601B (en) * | 2019-01-31 | 2021-09-14 | 浙江大学 | Design method of self-adaptive neural network synchronous robust controller of coordinated mechanical arm |
CN110427536A (en) * | 2019-08-12 | 2019-11-08 | 深圳忆海原识科技有限公司 | One type brain decision and kinetic control system |
CN112051733B (en) * | 2020-08-14 | 2021-08-10 | 华南理工大学 | Rigid mechanical arm composite learning control method based on image classification |
CN112051733A (en) * | 2020-08-14 | 2020-12-08 | 华南理工大学 | Rigid mechanical arm composite learning control method based on image classification |
CN112051734B (en) * | 2020-08-18 | 2021-07-23 | 华南理工大学 | Wheeled mobile robot event triggering tracking control method based on deterministic learning |
CN112051734A (en) * | 2020-08-18 | 2020-12-08 | 华南理工大学 | Wheeled mobile robot event triggering tracking control method based on deterministic learning |
CN112947093A (en) * | 2021-04-07 | 2021-06-11 | 长春工业大学 | Robot dispersion robust tracking control method, system and storage medium |
CN112947093B (en) * | 2021-04-07 | 2023-05-05 | 长春工业大学 | Distributed robust tracking control method, system and storage medium for robot |
CN113290563A (en) * | 2021-05-31 | 2021-08-24 | 哈尔滨工业大学 | Space target capturing method of variable-rigidity rope-driven snakelike mechanical arm |
CN113433825A (en) * | 2021-06-22 | 2021-09-24 | 广州大学 | Self-adaptive fault-tolerant control method and system of single-link mechanical arm and storage medium |
CN113433825B (en) * | 2021-06-22 | 2022-05-31 | 广州大学 | Self-adaptive fault-tolerant control method and system of single-link mechanical arm and storage medium |
CN117539153A (en) * | 2023-11-21 | 2024-02-09 | 山东大学 | Upper limb rehabilitation robot self-adaptive control method and system based on definite learning |
Also Published As
Publication number | Publication date |
---|---|
CN109227543B (en) | 2020-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109227543A (en) | A kind of limited pattern-based intelligent control method of flexible joint robot in position | |
CN108803316B (en) | Active fault-tolerant control method for faults of multi-agent system actuator | |
CN107160398B (en) | The safe and reliable control method of Rigid Robot Manipulator is limited based on the total state for determining study | |
CN109901403A (en) | A kind of face autonomous underwater robot neural network S control method | |
CN107498554B (en) | A kind of mechanical arm intelligent independent control method based on dynamic mode | |
CN102289204A (en) | Mechanical arm general control method based on determined learning theory | |
CN112947575B (en) | Unmanned aerial vehicle cluster multi-target searching method and system based on deep reinforcement learning | |
CN109799829B (en) | Robot group cooperative active sensing method based on self-organizing mapping | |
Li et al. | Application of bat algorithm based time optimal control in multi-robots formation reconfiguration | |
Caswara et al. | A neurofuzzy approach to the control of a flexible-link manipulator | |
CN112141369A (en) | Decision and control method for autonomous rendezvous and docking of spacecraft translation approach section | |
Liu et al. | Three-dimensional path following control of an underactuated robotic dolphin using deep reinforcement learning | |
Yang et al. | Control of space flexible manipulator using soft actor-critic and random network distillation | |
Flores et al. | Transition control of a tail-sitter UAV using recurrent neural networks | |
CN106569503A (en) | Triggering-type predictive control-based aircraft control method | |
Zhang et al. | Neural integrated control for a free-floating space robot with suddenly changing parameters | |
CN109933053B (en) | Unmanned aerial vehicle control method based on maneuvering action chain and unmanned aerial vehicle | |
Sadowska et al. | Formation control design for car-like nonholonomic robots using the backstepping approach | |
Zhang et al. | Multi-target encirclement with collision avoidance via deep reinforcement learning using relational graphs | |
Sohège et al. | A novel hybrid approach for fault-tolerant control of UAVs based on robust reinforcement learning | |
CN113156813B (en) | Pigeon swarm optimization control-based four-rotor unmanned aerial vehicle system with arms and control method | |
Xin et al. | Formation tracking and transformation control of nonholonomic AUVs based on improved SOM method | |
He et al. | Dynamic obstacle avoidance planning for manipulators of home | |
Liu et al. | Uniform synchronization in multi-axis motion control | |
Ge et al. | State-constrained bipartite tracking of interconnected robotic systems via hierarchical prescribed-performance 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 |