CN104238365A - Cantilever beam vibration control method on basis of self-adaption neural network control - Google Patents

Cantilever beam vibration control method on basis of self-adaption neural network control Download PDF

Info

Publication number
CN104238365A
CN104238365A CN201410500347.2A CN201410500347A CN104238365A CN 104238365 A CN104238365 A CN 104238365A CN 201410500347 A CN201410500347 A CN 201410500347A CN 104238365 A CN104238365 A CN 104238365A
Authority
CN
China
Prior art keywords
semi
girder
control
network
neural network
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
Application number
CN201410500347.2A
Other languages
Chinese (zh)
Inventor
胡桐月
费峻涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hohai University Changzhou Campus
Original Assignee
Hohai University Changzhou Campus
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hohai University Changzhou Campus filed Critical Hohai University Changzhou Campus
Priority to CN201410500347.2A priority Critical patent/CN104238365A/en
Publication of CN104238365A publication Critical patent/CN104238365A/en
Pending legal-status Critical Current

Links

Abstract

The invention discloses a cantilever beam vibration control method on the basis of self-adaption neural network control and is designed on the basis of filtered tracking error. A controller comprises proportional differential items and RBF neural network items. By the cantilever beam vibration control method, with unknown functions of a cantilever beam system of an RBF neural network approach, an updating algorithm of the weight of the RBF neural network is designed on the basis of Lyapunov stability theory, and overall stability of the system is guaranteed. Robust items are added into the updating algorithm, boundedness of control input is guaranteed, and the final tracking error is kept within any small range by the aid of proportional differential control items. Under the conditions of no structural or nonstructural parameters of a cantilever and with external interference, the control method is capable of accurately tracking and controlling the cantilever system, and robustness and reliability of the system are improved.

Description

Based on the semi-girder vibration control method of neural network control
Technical field
The present invention relates to a kind of semi-girder vibration control method based on neural network control, belong to semi-girder control technology field.
Background technology
To be one end of fingerboard be semi-girder does not produce the hold-down support of axis, perpendicular displacement and rotation, and the other end is free end (can produce and be parallel to power axial and perpendicular to axial direction).In engineering mechanics force analysis, more typical simplified model.In Practical Project is analyzed, most of Practical Project stressed member can be reduced to semi-girder.
Along with the development of science and technology is maked rapid progress, the develop rapidly of aeronautical and space technology, the expanding day of space operation scale, also more and more stricter to the requirement of aeronautical and space technology and space structure, a large amount of aerospace structures, the trend of flexibility development as all oriented in the space station of large modular, solar energy sailboard, satellite antenna, high-precision optical system and supporting mass structure, space manipulator etc.The use of flexible member not only increases the dirigibility of Spacecraft guidance and control and manufacture, also reduces launch cost simultaneously, and therefore, the extensive employing of flexible member is the trend of a certainty.But flexible member is Shortcomings part also, that is exactly that it easily produces elastic vibration when moving or locate, and at the end of motion, also can produce residual oscillation, caused vibration has very large impact to robust motion and positioning precision.Such as: robot for space and spacecraft flexible appendage as solar array etc. in disturbance cases, its significantly free vibration to continue for a long time, this will affect stability and point to control accuracy, especially when needs accurately control its position and point to.Similar accident was there is in international space in exploring, as the scopic rotating part of Landsat that the U.S. launches, vibrate owing to being subject to the interference of solar energy sailboard drive system, have impact on scopic steady operation, greatly reduce the quality that it transmits image, and for example " seeker No. l " communications satellite of U.S.'s transmitting in 1958, due to a lot of energy of dissipation of vibrations of its four whip antennas, after a period of time that normally worked, there is beyond thought satellite rollover phenomenon, finally cause mission failure.With regard to domestic, along with developing rapidly of China's Aero-Space cause, vibration problem also becomes increasingly conspicuous, as DFH-3 communications satellite just produced serious problem because of the vibration of solar energy sailboard.
Tradition suppresses the method for vibration to be by passive vibration isolation, and the method for vibration damping plays the object of vibration damping, and these Measures compare are passive, inflexible, bad adaptability, can not carry out real-time control to changeable extraneous vibration.In order to overcome above-mentioned difficulties, Active Vibration Control is suggested, and Active Vibration Control can experience extraneous Vibration Condition in real time, makes suitable reaction, exports control signal, suppresses vibration.The shock resistance of structural system for system functional reliability and precision is smoked to pass wants.A large amount of engineering structures generally bears the excitation of vibration environment in actual motion, is suppressed, will affect serviceability and the life-span of each parts in structure, can make its disabler time serious if do not take measures to the vibration of these structures.Therefore, in order to improve serviceability and the precision of structure, real-time vibration control must be carried out to structure.
International article has and is applied in the middle of the control of semi-girder by various advanced control method, typically has adaptive control and fuzzy control method.These achieve the TRAJECTORY CONTROL to semi-girder.But the robustness of adaptive control disturbance is to external world very low, system is easily made to become unstable.As can be seen here, semi-girder vibration control obviously still has inconvenience and defect, and is urgently further improved.
Summary of the invention
The object of the invention is to the defect overcoming the existence of existing semi-girder vibration control method, particularly improve cantilever beam system having that model is uncertain, under the various disturbed condition such as Parameter Perturbation and external disturbance power, to the tracking performance of ideal trajectory and the robustness of whole system, and provide a kind of semi-girder vibration control method based on neural network control.
The present invention solves its technical matters and realizes by the following technical solutions:
Based on the semi-girder vibration control method of neural network control, comprise the following steps:
1) the filtering error model based on semi-girder is set up;
2) CONTROLLER DESIGN;
3) adopt RBF network to approach the structure function of the semi-girder the unknown in semi-girder mathematical model, obtain the estimated value of the structure function of semi-girder the unknown;
4) based on the update algorithm of Lyapunov stability theory design RBF neural weights;
5) using the control inputs of the control inputs of the controller through RBF network control as semi-girder mathematical model, semi-girder is controlled, and real-time online upgrades.
Aforesaid step 1) in, the filtering error model based on semi-girder is:
Wherein, s is filtering error, and C is the damping term in semi-girder kinetic model, and u is the input vector in semi-girder kinetic model, and d is the disturbance of semi-girder kinetic model, and f (x) represents the structure function of semi-girder the unknown.
Aforesaid filtering error s is:
s = e . + Λe - - - ( 3 )
Wherein, Λ is filtering error parameter, and e is tracking error: e=q d-q
Q is semi-girder oscillation trajectory, q dfor the desirable oscillation trajectory of semi-girder.
The expression formula of structure function f (x) of aforesaid semi-girder the unknown is:
f ( x ) = q . . d = Λ e . + C ( q . d + Λe ) + Kq - - - ( 5 )
Wherein, K is the frequency item in semi-girder kinetic model,
for the signal that can measure,
Definition:
x = e T e . T q d T q . d T q . . d T T .
Aforesaid step 2) in, the control inputs of controller exported and proportion-plus-derivative control item by RBF network
Form:
Wherein, for RBF network exports, K vs is proportion-plus-derivative control item, K vfor linear Feedback Control parameter.
Aforesaid step 3) in, described RBF network is three-decker: input layer, hidden layer and output layer, described input layer in order to accept in system can measuring-signal input x, output after described hidden layer adopts Gaussian bases to calculate Nonlinear Mapping, described output layer obtains the output of whole RBF network by the output of each hidden node of weighting
The output of described RBF network is be the estimated value of the structure function of semi-girder the unknown,
Wherein for the estimated value of optimal network weights, φ (x) represents that hidden node exports.
Aforesaid step 4) in,
Described Lyapunov function V is chosen for:
V = 1 2 s T s + 1 2 tr ( W ~ T F - 1 W ~ ) - - - ( 17 )
Wherein, for network weight evaluated error, W *for optimal network weights,
The update algorithm of described RBF network weight is:
W ^ . = Fφ ( x ) s T - γF | | S | | W ^ - - - ( 16 )
Wherein, F=F t> 0 is the gain matrix of weighed value adjusting, γ be greater than 0 arbitrary value, be called forgetting factor.
In the update algorithm of aforesaid RBF network weight, meet:
| | s | | > γ W B 2 4 + ϵ b + b d K v min ≡ b s - - - ( 23 )
Or
| | W ~ | | F > W B 2 + W B 2 4 + ϵ b + b d γ ≡ b w - - - ( 24 )
Wherein, W brepresent optimal network weights W *the upper bound, K vminfor linear Feedback Control parameter K vsmallest real eigenvalue and be greater than zero, ε bfor the upper dividing value of network approximate error ε (x), b dfor the upper dividing value of disturbance d.
Aforesaid step 5) in, semi-girder mathematical model is:
q . . + C q . + Kq = u + d - - - ( 1 )
Wherein, C is damping term, and K is frequency item, and q is semi-girder oscillation trajectory, and u is input vector, and d is disturbance.
Compared with prior art, advantage is in the present invention:
(1) have employed the control method that self_adaptive RBF network and proportion-plus-derivative control combine, effectively can overcome unknown term and the external interference effect of Flexural cantilever model, greatly can improve tracking accuracy again.
(2) the present invention adopts neural network adaptive algorithm, can on-line control parameter control system, and adaptive algorithm, based on the design of Lyapunov stability theory, ensure that the global stability of closed-loop system.
(3) also add the robust item ensureing weights bounded in the adaptive algorithm of network weight, thus ensure that the boundedness of control inputs, make the present invention be easy to implement in engineering.
(4) the present invention does not need the control of semi-girder to be based upon on the basis of object Accurate Model, saves the expense of modeling.
Accompanying drawing explanation
Fig. 1 is the semi-girder vibration control principle composition based on neural network control of the present invention;
Fig. 2 is the track following figure after employing the present invention under Static disturbance.
Embodiment
For further setting forth the present invention for the technological means reaching predetermined goal of the invention and take and effect, below in conjunction with accompanying drawing and preferred embodiment, to the detailed description of the invention as rear.
A kind of semi-girder vibration control principle based on neural network control of the present invention as shown in Figure 1, comprises following components:
(1) set up based on semi-girder filtering error model
The vector form of semi-girder kinetic model:
q . . + C q . + Kq = u + d - - - ( 1 )
C, K ∈ R in formula i*ifor systematic parameter, wherein C is damping term, and K is frequency item, and q is semi-girder oscillation trajectory, and u is input vector, and d is disturbance.
The control objectives of cantilever beam system is that semi-girder oscillation trajectory follows the tracks of upper given ideal trajectory, if ideal trajectory q dfor: q d=[x d1, x d2..., x di] t, x difor ideal trajectory i-th order component, i represents exponent number.
Definition tracking error e is:
e=q d-q???(2)
Definition filtering error s expression formula is:
s = e . + Λe - - - ( 3 )
In formula, Λ=Λ t> 0 is filtering error parameter, and being generally namely taken as element is positive diagonal matrix entirely.Differentiate is carried out to filtering error s, obtains the filtering error model based on semi-girder:
s . = - Cs - u + f ( x ) - d - - - ( 4 )
In formula,
f ( x ) = q . . d = Λ e . + C ( q . d + Λe ) + Kq - - - ( 5 )
F (x) represents the structure function of semi-girder the unknown, in formula, for the signal that can measure, as the input of neural network, definition:
x = e T e . T q d T q . d T q . . d T T - - - ( 6 )
(2) CONTROLLER DESIGN
The control inputs of CONTROLLER DESIGN for:
In formula, for the estimated value of structure function f (x) of semi-girder the unknown, be the output of RBF network, the Nonlinear Mapping utilizing neural network powerful and approximation capability estimate its true value online in real time.
Kv is linear Feedback Control parameter, be proportion-plus-derivative control item, wherein K v = K v T > 0 .
By control inputs control inputs u as the mathematical model of semi-girder brings into (4), obtains closed-loop system equation to be:
s . = - Cs - f ^ ( x ) - K v s + f ( x ) - d = - ( K v + C ) s + f ~ ( x ) - d - - - ( 8 )
In formula, for the network-evaluated error of RBF:
f ~ = f - f ~ - - - ( 9 )
Formula (8) is the kinetics equation of filtered tracking error system, and the object of control system makes error s bounded, and converges on a less scope.From formula (3), be a stable wave filter from e to s, if s ultimate boundness, then tracking error e bounded.
For the RBF network in the present invention, select three-decker: input layer, hidden layer and output layer.Input layer accept in system can measuring-signal input x; Output after hidden layer adopts Gaussian bases to calculate Nonlinear Mapping; Output layer obtains the output of whole RBF network by the output of each hidden node of weighting, as follows with mathematical description RBF network model:
y i = Σ j = 1 n 2 ω ij φ j , i = 1,2 , . . . , n 3 - - - ( 10 )
φ j(x)=exp(||x-c j||/σ j),j=1,2,…n 2
In formula, n 2, n 3represent hidden node number and output layer node number respectively, and the dimension of input signal x is designated as n 1; ω ijrepresent network weight; y irepresent that RBF network exports; φ jx () is hidden node output; c j, σ jrepresent center vector and the sound stage width of each hidden node respectively.Existing document is verified, and RBF network can approach the nonlinear function of arbitrary smooth with arbitrary accuracy.Center vector and the sound stage width of the RBF network in the present invention are determined according to priori, are designed to fixed value, do not change in system operation, and the online real-time update of weights.Based on this, RBF network model is rewritten as:
y=W Tφ(x)???(11)
In formula, W t=[ω ij], φ (x)=[φ j(x)], due to c j, σ jfixing, φ (x) is known signal.
Based on the approximation capability of RBF network, can do to suppose as this: there is one group of optimal network weights W *, make when the input x of RBF network belong to one compact S time, RBF network can Nonlinear Function Approximation f (x), network approximate error ε (x) bounded under optimal network weights,
f(x)=W *Tφ(x)+ε(x)???(12)
In formula, || ε (x) ||≤ε b, ε bfor the upper bound of network approximate error ε (x);
Optimal network weights bounded: || W *|| f≤ W b, || || fthe F norm of representing matrix, W brepresent optimal network weights W *the upper bound.
Utilize RBF network to approach f (x), obtain the estimated value of nonlinear function f (x)
f ^ ( x ) = W ^ T φ ( x ) - - - ( 13 )
for optimal network weights W *estimated value.
Convolution (7), the control inputs of controller becomes:
By control inputs control inputs u as the mathematical model of semi-girder brings formula (4) into:
s . = - ( K v + C ) s + f ( x ) - f ^ ( x ) - d = - ( K v + C ) s + W T φ ( x ) + ϵ ( x ) - W ^ T φ ( x ) - d = - ( K v + C ) s + W ~ T φ ( x ) + ϵ ( x ) - d - - - ( 15 )
In formula, for network weight evaluated error.
So far, structure and the closed-loop error equation of the controller described in invention is obtained.
(3) update algorithm of RBF neural weights is designed
Semi-girder vibration control method based on neural network control of the present invention, the update algorithm of its RBF network weight is:
W ^ = Fφ ( x ) s T - γF | | s | | W ^ - - - ( 16 ) .
In formula, F=F t> 0 is the arbitrary value of the gain matrix of weighed value adjusting, γ > 0, is called forgetting factor.
Prove that the network weight update algorithm in formula (16) can ensure tracking error s and network weight estimated value below ultimate boundness, and respective boundary is as shown in lower inequality (24), (25) right side.Meanwhile, by increasing linear Feedback Control parameter K v, tracking error s can be made to be maintained to arbitrarily small scope.
A Lyapunov candidate functions V is chosen to the closed-loop system of formula (15):
V = 1 2 s T s + 1 2 tr ( W ~ T F - 1 W ~ ) - - - ( 17 )
To formula (17) both sides differentiate:
V . = s T s . = tr ( W ~ T F - 1 W ~ . ) = - s T ( K v + C ) s + tr ( W ~ T ( F - 1 W ~ . + φ ( x ) s T ) ) + s T ( ϵ ( x ) - d ) - - - ( 18 )
Because W ~ = W * - W ^ , So W ~ = - W ^ . .
Network weight update algorithm in formula (16) is brought into formula (18) to obtain:
V . = - s T ( K v + C ) s + γ | | s | | tr { W ~ T ( W - W ~ ) } + s T ( δ ( x ) - d ) - - - ( 19 )
The character of associate(d) matrix mark and matrix F norm, has:
tr { W ~ T ( W - W ~ ) } = ⟨ W ~ , W ⟩ F - | | W ~ | | F 2 ≤ | | W ~ | | F | | W | | F - | | W ~ | | F 2 - - - ( 20 )
So,
V . ≤ - s T K v s + γ | | s | | | | W ~ | | F ( | | W | | F - | | W ~ | | F ) + ( ϵ b + b d ) | | s | | ≤ - K v min | | s | | + γ | | s | | | | W ~ | | F ( W B - | | W ~ | | F ) + ( ϵ b + b d ) | | s | | = - | | s | | [ K v min | | s | | + γ | | W ~ | | F ( | | W ~ | | F - W B ) - ( ϵ b + d d ) ] - - - ( 21 )
Wherein, K vminfor linear Feedback Control parameter K vsmallest real eigenvalue and be greater than zero, ε bfor the upper dividing value of network approximate error ε (x), b dfor the upper dividing value of disturbance d, namely || ε (x) ||≤ε b, || d||≤b d.
As can be seen from formula (21), be just such as formula transition formula evaluation in middle bracket, then be negative, i.e. system stability.
Due to:
K v min | | s | | + γ | | W ~ | | F ( | | W ~ | | F - W B ) - ( ϵ b + b d ) = γ ( | | W ~ | | F - W B 2 ) 2 - γ W B 2 4 + K v min | | s | | - ( ϵ b + b d ) - - - ( 22 )
It is just that formula (22) is ensured, as long as
| | s | | > γ W B 2 4 + ϵ b + b d K v min ≡ b s - - - ( 23 )
Or
| | W ~ | | F > W B 2 + W B 2 4 + ϵ b + b d γ ≡ b w - - - ( 24 )
In formula, b s, b wa just constant of arbitrarily setting.
Comprehensive above analytic process, can reach a conclusion: negative definite outside the region that formula (23) or (24) describe.According to the standard extension theorem of Lyapunov stability theory, known || s|| and ultimate boundedness be guaranteed.Because, once s or beyond the regional extent that formula (23) and (24) specify, the decline of Lyapunov function V will be caused, this can make again s and in the scope that two formulas of getting back to limit.So formula (23) and formula (24) have in fact respectively provided || s|| and the upper bound.And, can notice from formula (23), larger feedback gain matrix K vwill obtain less || the upper bound of s||, therefore tracking error s finally can maintain in arbitrarily small scope.As can be seen from formula (24), the boundedness of RBF network weight, ensure that the boundedness of control inputs, and this point is very important for Practical Project.
As can be seen from analytical proof process above, Section 1 in right value update algorithm in formula (16) is the error back propagation algorithm derived based on lyapunov stability theory, and Section 2 is the robust item added, be used for ensureing the boundedness of weights.
(4) computer simulation experiment
By dynamics, the main contribution of structural vibration in minimum several rank mode, in order to explain the situation and fundamental purpose in the emulation of elastic construction vibration suppression, only select first step mode here.In order to show the validity of the semi-girder vibration control method based on neural network control that the present invention proposes more intuitively, perceptive construction on mathematics/SIMULINK is now utilized to carry out computer simulation experiment to this control program.With reference to existing document, choose:
The parameter of semi-girder is: damping term C=0.18, frequency item K=56.4.
Desired reference track is set as: q d = 0 , q . d = 0 , q . . d = 0 .
Linear Feedback Control parameter is taken as K v=100, filtering error parameter is taken as Λ=1.
The node in hidden layer n of RBF neural 2be taken as 45.
External interference added by emulation experiment is white noise disturbance d=randn (1,1), if disturbance applies all the time, the 10th second time, apply control action Dynamic simulation program, obtain the experimental result of the neural network control system of semi-girder as shown in Figure 2.
Fig. 2 illustrates the semi-girder oscillation trajectory tracking effect curve under the control method proposed in the present invention, and in figure, solid line is ideal trajectory, and dotted line is actual path.As can be seen from this accompanying drawing, under stable state disturbance, neural network control effect clearly, and after applying control action, external interference decays rapidly in 3s ~ 4s.Control system can make the output of semi-girder, when not knowing semi-girder parameter and structure and there is external interference effect, can promptly follow the tracks of given ideal trajectory, and tracking error is very little, reaches satisfied effect.
As can be seen from above analogous diagram, the control method that the present invention proposes is followed the tracks of the oscillation trajectory of semi-girder good control effects, substantially increase tracking performance and the robustness of cantilever beam system, control to provide theoretical foundation and From Math to the high precision of semi-girder oscillation trajectory.
The content be not described in detail in instructions of the present invention belongs to the known technical know-how of professional and technical personnel in the field.
The above, it is only preferred embodiment of the present invention, not any in form large restriction is done to the present invention, although the present invention discloses as above with preferred embodiment, but and be not used to limit the present invention, any those skilled in the art, do not departing within the scope of technical solution of the present invention, make a little change when the technology contents of above-mentioned announcement can be utilized or be modified to the Equivalent embodiments of equivalent variations, in every case be the content not departing from technical solution of the present invention, according to any simple modification that technical spirit of the present invention is done above embodiment, equivalent variations and modification, all still belong in the scope of our bright technical scheme.

Claims (9)

1., based on the semi-girder vibration control method of neural network control, it is characterized in that, comprise the following steps:
1) the filtering error model based on semi-girder is set up;
2) CONTROLLER DESIGN;
3) adopt RBF network to approach the structure function of the semi-girder the unknown in semi-girder mathematical model, obtain the estimated value of the structure function of semi-girder the unknown;
4) based on the update algorithm of Lyapunov stability theory design RBF neural weights;
5) using the control inputs of the control inputs of the controller through RBF network control as semi-girder mathematical model, semi-girder is controlled, and real-time online upgrades.
2. the semi-girder vibration control method based on neural network control according to claim 1, is characterized in that, described step 1) in, the filtering error model based on semi-girder is:
Wherein, s is filtering error, and C is the damping term in semi-girder kinetic model, and u is the input vector in semi-girder kinetic model, and d is the disturbance of semi-girder kinetic model, and f (x) represents the structure function of semi-girder the unknown.
3. the semi-girder vibration control method based on neural network control according to claim 2, is characterized in that, described filtering error s is:
s = e . + Λe - - - ( 3 )
Wherein, Λ is filtering error parameter, and e is tracking error: e=q d-q
Q is semi-girder oscillation trajectory, q dfor the desirable oscillation trajectory of semi-girder.
4. the semi-girder vibration control method based on neural network control according to claim 2, is characterized in that, the expression formula of structure function f (x) of described semi-girder the unknown is:
f ( x ) = q . . d = Λ e . + C ( q . d + Λe ) + Kq - - - ( 5 )
Wherein, K is the frequency item in semi-girder kinetic model,
for the signal that can measure,
Definition:
x = e T e . T q d T q . d T q . . d T T .
5. the semi-girder vibration control method based on neural network control according to claim 1, is characterized in that, described step 2) in, the control inputs of controller exported by RBF network and form with proportion-plus-derivative control item:
Wherein, for RBF network exports, K vs is proportion-plus-derivative control item, K vfor linear Feedback Control parameter.
6. the semi-girder vibration control method based on neural network control according to claim 1, it is characterized in that, described step 3) in, described RBF network is three-decker: input layer, hidden layer and output layer, described input layer in order to accept in system can measuring-signal input x, the output after described hidden layer adopts Gaussian bases to calculate Nonlinear Mapping, described output layer obtains the output of whole RBF network by the output of each hidden node of weighting
The output of described RBF network is be the estimated value of the structure function of semi-girder the unknown,
Wherein for the estimated value of optimal network weights, φ (x) represents that hidden node exports.
7. the semi-girder vibration control method based on neural network control according to claim 1, is characterized in that, described step 4) in,
Described Lyapunov function V is chosen for:
V = 1 2 s T s + 1 2 tr ( W ~ T F - 1 W ~ ) - - - ( 17 )
Wherein, for network weight evaluated error, W *for optimal network weights,
The update algorithm of described RBF network weight is:
W ^ . = Fφ ( x ) s T - γF | | S | | W ^ - - - ( 16 )
Wherein, F=F t> 0 is the gain matrix of weighed value adjusting, γ be greater than 0 arbitrary value, be called forgetting factor.
8. the semi-girder vibration control method based on neural network control according to claim 7, is characterized in that, in the update algorithm of described RBF network weight, meets:
| | s | | > γ W B 2 4 + ϵ b + b d K v min ≡ b s - - - ( 23 )
Or
| | W ~ | | F > W B 2 + W B 2 4 + ϵ b + b d γ ≡ b w - - - ( 24 )
Wherein, W brepresent optimal network weights W *the upper bound, K vminfor linear Feedback Control parameter K vsmallest real eigenvalue and be greater than zero, ε bfor the upper dividing value of network approximate error ε (x), b dfor the upper dividing value of disturbance d.
9. the semi-girder vibration control method based on neural network control according to claim 1, is characterized in that, described step 5) in, semi-girder mathematical model is:
q . . + C q . + Kq = u + d - - - ( 1 )
Wherein, C is damping term, and K is frequency item, and q is semi-girder oscillation trajectory, and u is input vector, and d is disturbance.
CN201410500347.2A 2014-09-25 2014-09-25 Cantilever beam vibration control method on basis of self-adaption neural network control Pending CN104238365A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410500347.2A CN104238365A (en) 2014-09-25 2014-09-25 Cantilever beam vibration control method on basis of self-adaption neural network control

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410500347.2A CN104238365A (en) 2014-09-25 2014-09-25 Cantilever beam vibration control method on basis of self-adaption neural network control

Publications (1)

Publication Number Publication Date
CN104238365A true CN104238365A (en) 2014-12-24

Family

ID=52226685

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410500347.2A Pending CN104238365A (en) 2014-09-25 2014-09-25 Cantilever beam vibration control method on basis of self-adaption neural network control

Country Status (1)

Country Link
CN (1) CN104238365A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104808487A (en) * 2015-03-03 2015-07-29 台州学院 Neural network adaptive robust trajectory tracking method and controller
CN104950674A (en) * 2015-06-12 2015-09-30 河海大学常州校区 Cantilever beam vibration control method based on back-stepping fuzzy sliding mode control
CN105157727A (en) * 2015-09-08 2015-12-16 河海大学常州校区 Control method for neural network global sliding mode of micro gyroscope based on linearization feedback
CN106292278A (en) * 2016-08-18 2017-01-04 大连理工大学 The cantilever beam control method that a kind of many piezoelectric fibre composite materials drive
CN106773705A (en) * 2017-01-06 2017-05-31 西安交通大学 A kind of adaptive active control method and active control system for vibration damping de-noising
CN106875008A (en) * 2017-03-02 2017-06-20 广东工业大学 A kind of Inverse Solution For Manipulator Kinematics method and system
CN108398885A (en) * 2018-03-29 2018-08-14 湖南大学 Rotor flying mechanical arm self_adaptive RBF NNs surveys Auto-disturbance-rejection Control of making an uproar

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH03245065A (en) * 1990-02-23 1991-10-31 Hitachi Ltd Fourier transform neural network apparatus
JPH09245012A (en) * 1996-03-08 1997-09-19 Mitsubishi Electric Corp Learning control method
CN101055219A (en) * 2007-05-11 2007-10-17 上海大学 Simulated intelligence flexible space sail board structural vibration main control test platform and method
CN101299004A (en) * 2008-06-24 2008-11-05 华南理工大学 Vibrating failure diagnosis method based on determined learning theory
CN101763035A (en) * 2009-11-13 2010-06-30 上海电力学院 Method for controlling radial basis function (RBF) neural network tuned proportion integration differentiation (PID) and fuzzy immunization
CN102073276A (en) * 2011-02-21 2011-05-25 哈尔滨工业大学 Method for controlling flexible structure and self-adaptive changing structure by radial basis function (RBF) neural network
CN102289204A (en) * 2011-06-03 2011-12-21 华南理工大学 Mechanical arm general control method based on determined learning theory
CN102436176A (en) * 2011-10-20 2012-05-02 河海大学常州校区 Micro-gyroscope control system based on neural network
CN102914972A (en) * 2012-11-01 2013-02-06 河海大学常州校区 Micro-gyroscope RBF (Radial Basis Function) network self-adapting control method based on model global approximation
CN103869702A (en) * 2014-03-13 2014-06-18 大连理工大学 Active chattering control method for flexible cantilever beam structure

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH03245065A (en) * 1990-02-23 1991-10-31 Hitachi Ltd Fourier transform neural network apparatus
JPH09245012A (en) * 1996-03-08 1997-09-19 Mitsubishi Electric Corp Learning control method
CN101055219A (en) * 2007-05-11 2007-10-17 上海大学 Simulated intelligence flexible space sail board structural vibration main control test platform and method
CN101299004A (en) * 2008-06-24 2008-11-05 华南理工大学 Vibrating failure diagnosis method based on determined learning theory
CN101763035A (en) * 2009-11-13 2010-06-30 上海电力学院 Method for controlling radial basis function (RBF) neural network tuned proportion integration differentiation (PID) and fuzzy immunization
CN102073276A (en) * 2011-02-21 2011-05-25 哈尔滨工业大学 Method for controlling flexible structure and self-adaptive changing structure by radial basis function (RBF) neural network
CN102289204A (en) * 2011-06-03 2011-12-21 华南理工大学 Mechanical arm general control method based on determined learning theory
CN102436176A (en) * 2011-10-20 2012-05-02 河海大学常州校区 Micro-gyroscope control system based on neural network
CN102914972A (en) * 2012-11-01 2013-02-06 河海大学常州校区 Micro-gyroscope RBF (Radial Basis Function) network self-adapting control method based on model global approximation
CN103869702A (en) * 2014-03-13 2014-06-18 大连理工大学 Active chattering control method for flexible cantilever beam structure

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
C. MAKKAR 等: "Lyapunov-Based Tracking Control in the Presence of Uncertain Nonlinear Parameterizable Friction", 《IEEE TRANSACTIONS ON AUTOMATIC CONTROL》 *
柴华: "机械臂操作柔性负载系统的振动控制方法", 《中国优秀学位论文全文数据库 信息科技辑》 *
王学林 等: "基于卡尔曼滤波的RBF神经网络和PD复合控制研究", 《计算机测量与控制》 *
王平 等: "《物联网概论》", 31 January 2014, 北京大学出版社 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104808487A (en) * 2015-03-03 2015-07-29 台州学院 Neural network adaptive robust trajectory tracking method and controller
CN104950674A (en) * 2015-06-12 2015-09-30 河海大学常州校区 Cantilever beam vibration control method based on back-stepping fuzzy sliding mode control
CN105157727A (en) * 2015-09-08 2015-12-16 河海大学常州校区 Control method for neural network global sliding mode of micro gyroscope based on linearization feedback
CN105157727B (en) * 2015-09-08 2018-11-23 河海大学常州校区 Gyroscope neural network total-sliding-mode control method based on Linearization Feedback
CN106292278A (en) * 2016-08-18 2017-01-04 大连理工大学 The cantilever beam control method that a kind of many piezoelectric fibre composite materials drive
CN106292278B (en) * 2016-08-18 2019-03-05 大连理工大学 A kind of cantilever beam control method of more piezoelectric fibre composite material drivings
CN106773705A (en) * 2017-01-06 2017-05-31 西安交通大学 A kind of adaptive active control method and active control system for vibration damping de-noising
CN106773705B (en) * 2017-01-06 2018-10-19 西安交通大学 A kind of adaptive active control method and active control system for vibration damping de-noising
CN106875008A (en) * 2017-03-02 2017-06-20 广东工业大学 A kind of Inverse Solution For Manipulator Kinematics method and system
CN108398885A (en) * 2018-03-29 2018-08-14 湖南大学 Rotor flying mechanical arm self_adaptive RBF NNs surveys Auto-disturbance-rejection Control of making an uproar

Similar Documents

Publication Publication Date Title
CN104238365A (en) Cantilever beam vibration control method on basis of self-adaption neural network control
Hu et al. Robust saturated finite time output feedback attitude stabilization for rigid spacecraft
CN103616818B (en) The neural overall fast terminal sliding-mode control of adaptive fuzzy of gyroscope
CN104950674A (en) Cantilever beam vibration control method based on back-stepping fuzzy sliding mode control
Fei et al. Adaptive control of MEMS gyroscope using global fast terminal sliding mode control and fuzzy-neural-network
Hu et al. Sliding mode and shaped input vibration control of flexible systems
CN104122794B (en) The adaptive fuzzy nerve compensation non-singular terminal sliding-mode control of gyroscope
Wang et al. Robust adaptive sliding mode control of MEMS gyroscope using T–S fuzzy model
Hu et al. Intelligent proportional-derivative control for flexible spacecraft attitude stabilization with unknown input saturation
CN102914972A (en) Micro-gyroscope RBF (Radial Basis Function) network self-adapting control method based on model global approximation
CN103116275A (en) Robust neural network control system for micro-electro-mechanical system (MEMS) gyroscope based on sliding mode compensation and control method of control system
CN103529701A (en) Method of global sliding mode control of neural network of micro-gyroscope
Dang et al. Formation control using μ-synthesis for inner-formation gravity measurement satellite system
Qin et al. Robust parameter dependent receding horizon h∞ control of flexible air‐breathing hypersonic vehicles with input constraints
Zhang et al. A novel underactuated control scheme for deployment/retrieval of space tethered system
He et al. Active vibration control and stability analysis of flexible beam systems
Xie et al. A data driven control method for structure vibration suppression
Hu et al. Output feedback attitude tracking for spacecraft under control saturation and disturbance
An et al. Sliding mode differentiator based tracking control of uncertain nonlinear systems with application to hypersonic flight
CN102866633B (en) Dynamic sliding-mode control system of miniature gyroscope
Zhang et al. Analysis of Reflector Vibration-Induced Pointing Errors for Large Antennas Subject to Wind Disturbance: Evaluating the pointing error caused by reflector deformation.
Cai et al. Command tracking control system design and evaluation for hypersonic reentry vehicles driven by a reaction control system
Vigano et al. Optimal periodic output feedback control: a continuous-time approach and a case study
Cao et al. Are phase and time-delay margins always adversely affected by high-gain?
CN105717791A (en) Cantilever beam vibration control method adaptive to H-infinite control

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20141224