CN106200383B  A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network  Google Patents
A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network Download PDFInfo
 Publication number
 CN106200383B CN106200383B CN201610645413.4A CN201610645413A CN106200383B CN 106200383 B CN106200383 B CN 106200383B CN 201610645413 A CN201610645413 A CN 201610645413A CN 106200383 B CN106200383 B CN 106200383B
 Authority
 CN
 China
 Prior art keywords
 neural network
 control
 frame
 adaptive neural
 interference
 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.)
 Active
Links
 230000003044 adaptive Effects 0.000 title claims abstract description 68
 230000001537 neural Effects 0.000 title claims abstract description 66
 239000011159 matrix material Substances 0.000 claims abstract description 43
 230000000875 corresponding Effects 0.000 claims description 24
 238000005259 measurement Methods 0.000 claims description 7
 230000001276 controlling effect Effects 0.000 claims description 3
 230000002452 interceptive Effects 0.000 claims description 3
 241000208340 Araliaceae Species 0.000 claims 1
 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
 235000005035 ginseng Nutrition 0.000 claims 1
 235000008434 ginseng Nutrition 0.000 claims 1
 230000000694 effects Effects 0.000 description 4
 230000003287 optical Effects 0.000 description 2
 239000000969 carrier Substances 0.000 description 1
 238000000034 method Methods 0.000 description 1
 238000009966 trimming Methods 0.000 description 1
Classifications

 G—PHYSICS
 G05—CONTROLLING; REGULATING
 G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
 G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
 G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
 G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
 G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Abstract
A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network, it is related to the Disturbance Rejection design of control parameter Online Estimation and extended state observer based on adaptive neural network, first, according to three axis inertially stabilized platform kinetic models, for the uncertain caused feedback control parameters timevarying characteristics of three axis inertially stabilized platform kinetic parameters, it constructs adaptive neural network and Online Estimation is carried out to feedback control matrix parameter, three axis inertially stabilized platforms control precision is made to approach expectational model control precision；Secondly, influence for adaptive neural network evaluated error and the interference compensation control amount of the upper bound building of three axis inertially stabilized platforms interference to control precision, building extended state observer is estimated and is inhibited to interference, realizes three axis inertially stabilized platform highprecision controls under complex environment.The present invention has many advantages, such as that realtime is good, dynamic parameter response is fast, it is adaptable to interfere multisource, can be used for highprecision control etc. of the three axis inertially stabilized platforms under complicated multisource interference environment.
Description
Technical field
The three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network that the present invention relates to a kind of are fitted
For aerial mapping stabilized platform highprecision control field.
Background technique
Three axis inertially stabilized platforms are fixed on flight carrier by pedestal, support and stablize remotely sensed image load, be isolated more
The influence to the remotely sensed image load optical axis is interfered in source, is improved remote sensing load pointing accuracy over the ground, is with a wide range of applications.
Three axis inertially stabilized platforms interference type multiplicity during the work time, do not disturb there is only uncertain RANDOM WIND and
Caused angular movement interference is vibrated by aircraft body, there is also due to load system mass center be not overlapped with system origin it is caused not
Trimming moment, internal interference caused by platform sensor part measurement error, therefore, the complicated highprecision control sides disturbed under environment more
Method has become one of the key technology of three axis inertially stabilized platform researchs.
To improve performance, PID control method, robust control, intelligent control method are used for three axis inertially stabilized platforms
Highprecision control.PID controller structure is simple, but poor anti jamming capability, it is difficult to ensure that three axis inertia under multisource interference effect
Platform stable precision.Robust control can inhibit system model parameter uncertainty and multisource to interfere the influence to control precision,
But controlling precision has biggish conservative.By a large amount of sample training, neural network infinite can approach nonlinear system,
To solve the nonlinear of three axis inertially stabilized platform kinetic models, highprecision gesture stability, but traditional nerve are realized
Network needs a large amount of sample data to be trained, and has the shortcomings that realtime is poor.
Summary of the invention
Technical problem solved by the present invention is three axis inertially stabilized platforms control precision by system model is nonlinear and multisource
The influence of interference solves the nonlinear of model by adaptive neural network realtime estimation feedback control parameters, and constructs expansion
Observer improves the antiinterference ability of system, realizes three axis inertially stabilized platform highprecision controls under complex environment.
The technical solution of invention are as follows: kinetic model is established first against three axis inertially stabilized platforms, according to expectation
System performance index constructs three axis inertially stabilized platform reference driving force models, is fed back and is controlled in line tracking by adaptive network
Parameter processed approaches three axis inertially stabilized platform kinetic models and reference model；Secondly, building expansion observer, reduces certainly
Adapt to influence of the sign function gain of the upper bound building of neural network evaluated error and external disturbance to system control precision.Its
Realize that steps are as follows:
(1) according to three axis inertially stabilized platform kinetic models, design reference model and feedback controller, feedback control is solved
Parameter processedIt is special for the uncertain caused feedback control parameters timevarying of three axis inertially stabilized platform kinetic parameters
Property, the feedback control parameters of adaptive neural network Online Estimation timevariant nonlinear function are constructed,
Wherein,It is the weight matrix of adaptive neural network,It is adaptive neural network
Basic function, l are the number of nodes of adaptive neural network hidden layer,For the state variable of system, n=6 is state variable
Dimension, ε_{j}It is the approximate error of adaptive neural network jth row, when j=1 indicates roll frame, j=2 pitching frame, table when j=3
Show orientation frame, three axis inertially stabilized platforms control precision is made to approach expectational model control precision；
(2) interference for the upper bound building interfered for adaptive neural network evaluated error and three axis inertially stabilized platforms is mended
Influence of the control amount to system control precision is repaid, building extended state observer is estimated and is inhibited to interference,
Wherein,WithRespectively state variable z_{1}And z_{2}Estimated value, z_{1}=e, z_{2}=Δ=B ε x+Dd, e is current shape
The state variable x of state amount x and reference model^{mm}Difference,It is adaptive neural network weight square realtime estimation value,
For the coefficient matrix of frame input, m=3 is the dimension of system input, and ε is the approximate error of adaptive neural network, and d is system
Frame interference, For the state equation of reference model
Matrix,WithIt is that POLE PLACEMENT USING principle is based on according to dynamic performance requirement
The matrix artificially designed, r are interference compensation control amount, w_{0}> 0 is design variable, realizes three axis stable inertias under complex environment
Platform highprecision control；
The three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network of the invention, wherein described
Step (1) is for the uncertain caused feedback control parameters timevarying characteristics of three axis inertially stabilized platform kinetic parameters, structure
Build the feedback control parameters of adaptive neural network Online Estimation timevariant nonlinear functionThree axis inertially stabilized platforms are moved
Mechanical model is expressed as
Wherein, x=[θ_{j} ω_{j}]^{T},H=0_{3×3}, F=(f_{jk}), j, k=1,2,3,
U=[u_{1} u_{2} u_{3}]^{T}, g=[g_{1} g_{2} g_{3}]^{T}, d=[d_{1} d_{2} d_{3}]^{T},
Wherein, θ_{j}For corresponding j frame angle, ω_{j}For corresponding j frame angular speed,What it is for state variable is
Matrix number, u_{j}For the input of corresponding j frame voltage, g is the estimated value of the gimbal moment of intertia, d_{j}For the interference of system j frame, by
The perturbation value △ g of system interference, the gimbal moment of intertia_{j}With the control input perturbation △ b as caused by measurement noise_{j}It constitutes, b_{j}For frame
The ideal control coefrficient of the corresponding j column of the coefficient matrix B of frame input, F are the corresponding perfect condition variable system with timevarying characteristics
Matrix number；
Design reference model
WhereinIt is the control input of desired reference model；
Design of feedback controller
Wherein r=[r_{1} r_{2} r_{3}]^{T}It is the control amount interfering d for compensation and designing, meets It is feedback control parameters；
Then system
Feedback control parameters
Due to h_{jk},f_{jk}, k, j=1,2,3 be timevariant nonlinear function,It is also timevariant nonlinear function, therefore, utilizes
The infinite extensive approximation capability of adaptive neural network carrys out Online Estimation timevariant nonlinear function
Control based on model reference adaptive ANN Control inputs u_{j}, adaptive neural network turnover rateWith
Interference compensation control amount r_{j}Expression formula be respectively
r_{j}=δ_{j}sgn(e^{T}PB_{*j})
Wherein, B_{*j}For the jth column of the coefficient matrix B of frame input, matrix P is the positive definite symmetric solution of state equation,
PA^{mm}+A^{mmT}P=Q
WhereinIt is positive definite symmetrical matrix,It is adaptive law gain matrix, adaptive neural network is estimated
The upper bound of meter error and external disturbance is
δ_{j}>ε_{j}x+d_{j}/b_{j}
Wherein, ε_{j}It is the approximate error of adaptive neural network；
The three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network of the invention, wherein described
Step (2) building extended state observer is estimated and is inhibited to interference, realizes that three axis stable inertias under complex environment are flat
Platform highprecision control, three axis inertially stabilized platform error state equations are
Wherein,S is the rate of change of Δ, and the extended state observer of building is,
Revised interference compensation control amount r_{j}Expression formula be
Wherein, b_{j}For the ideal control coefrficient of the coefficient matrix B corresponding j column of frame input,It isJth+3
A component,Residual error B ε x and system interference Dd and disturbance estimated value z are estimated for system selfadaption neural network_{2}It is residual
Differenced_{j}For the interference of system j frame, by the perturbation value △ of system interference, the gimbal moment of intertia
g_{j}With the control input perturbation △ b as caused by measurement noise_{j}It constitutes, ε_{j}It is the approximate error of adaptive neural network, table when j=1
Show roll frame, j=2 pitching frame, when j=3 indicates orientation frame, B_{*j}For the jth column of the coefficient matrix B of frame input, matrix P is shape
The positive definite symmetric solution of state equation,
PA^{mm}+A^{mmT}P=Q
WhereinIt is positive definite symmetrical matrix；
Corresponding system, which inputs, is
Wherein,It is the corresponding j column control input of desired reference model, g_{j}For the estimation of the corresponding j gimbal moment of intertia
Value.
The advantages of the present invention over the prior art are that:
(1) present invention passes through the infinite extensive approximation capability of adaptive neural network, the feedback control of realtime estimation system
Parameter, solve the problems, such as due to model parameter it is nonlinear, it is uncertain caused by feedback control parameters timevariant nonlinears, make practical mould
Type approaches reference model, and control method structure is simple, strong antijamming capability；
(2) present invention is further estimated using expansion observer in the case where adaptive neural network guarantees that system is stablized
Disturbance suffered in the three axis inertially stabilized platform courses of work is counted and inhibits, control precision is high, can satisfy three axis stable inertias
The highprecision control demand of platform；
(3) present invention only requires according to the status information in the three axis inertially stabilized platform courses of work, Li Yapunuo is utilized
Husband's function designs adaptive neural network right value update matrix, can not be needed with the weight of online updating adaptive neural network
Any sample training has the advantages that data acquisition is convenient, it is simple to calculate.
Detailed description of the invention
Fig. 1 is three axis inertially stabilized platform control flows；
Fig. 2 is three axis inertially stabilized platform pitch channel control effects in flight experiment；
Fig. 3 is three axis inertially stabilized platform roll channel control effects in flight experiment；
Fig. 4 is three axis inertially stabilized platform azimuthal channel control effects in flight experiment.
Specific embodiment
As shown in Figure 1, of the invention is implemented as follows
(1) building is based on model reference adaptive neural network
Based on NewtonEuler equation, the kinetics equation of three axis inertially stabilized platforms is expressed as
Wherein, x=[θ_{j} ω_{j}]^{T},H=0_{3×3}, F=(f_{jk}), j, k=1,2,3,
U=[u_{1} u_{2} u_{3}]^{T},G=
[g_{1} g_{2} g_{3}]^{T}, d=[d_{1} d_{2} d_{3}]^{T},
Wherein, indicate that roll frame, j=2 pitching frame indicate orientation frame when j=3 when j=1,For the state of system
Variable, n=6 are the dimension of state variable, θ_{j}For corresponding j frame angle, ω_{j}For corresponding j frame angular speed,
For the coefficient matrix of state variable,For the coefficient matrix of frame input, m=3 is the dimension of system input, u_{j}For phase
The j frame voltage input answered, g are the estimated value of the gimbal moment of intertia, and d is system framework interference, by system interference, frame member
The perturbation value △ g of inertia_{j}With the control input perturbation △ b as caused by measurement noise_{j}It constitutes, b_{j}For the coefficient matrix B of frame input
The ideal control coefrficient of corresponding j column, F are corresponding perfect condition variation coefficient matrix, wherein
f_{31}=f_{32}=0,
Wherein, N indicates motor drive ratio, K_{e}Indicate back electromotive force constant, K_{t}Indicate motor torque coefficient, R_{m}Indicate motor
Resistance, J_{m}Indicate the rotary inertia of motor, J_{a}=diag (J_{ax},J_{ay},J_{az}) it is the rotary inertia of orientation frame in orientation frame coordinate system
Projection on the direction x, y, z, J_{p}=diag (J_{px},J_{py},J_{pz}) it is the rotary inertia of pitching frame in pitching frame coordinate system x, y, the side z
Upward projection, J_{r}=diag (J_{rx},J_{ry},J_{rz}) be roll frame rotary inertia in roll frame in the direction roll coordinate system x, y, z
On projection,For k system relative inertness space angular speed in k system x, y, the side z
Downward mapping；Respectively roll frame opposite base exists in roll coordinate system, pitching frame with respect to roll frame
The angular speed of pitching coordinate system, orientation frame with respect to pitching frame in azimuthal coordinates system, θ_{r}Indicate roll frame opposite base corner, θ_{p}Table
Show corner of the pitching frame with respect to roll frame, θ_{a}Corner of the orientation frame with respect to pitching frame is indicated, by the codedisc being mounted on gimbal axis
Measurement obtains, T_{dm}To act on the disturbance torque on motor, T_{dj}For the disturbance torque acted on j frame, rotary inertia is
J_{1}=J_{az}+N^{2}J_{m},
J_{2}=J_{px}+J_{ax}+N^{2}J_{m},
J_{3}=J_{ry}+(J_{ay}+J_{py})cos^{2}θ_{p}+(J_{az}+J_{pz})sin^{2}θ_{p}+N^{2}J_{m}
Design reference model
WhereinFor the state equation matrix of reference model,
WithIt is the matrix artificially designed according to dynamic performance requirement based on POLE PLACEMENT USING principle,It is the control input of desired reference model,
Design of feedback controller
Wherein r=[r_{1} r_{2} r_{3}]^{T}It is the control amount interfering d for compensation and designing, meets It is feedback control parameters, then system
IfThen real system is infinite approaches desired reference model；
Therefore, feedback control parameters
Due to h_{jk},f_{jk}, k, j=1,2,3 be timevariant nonlinear function,It is also timevariant nonlinear function, therefore, benefit
With the infinite extensive approximation capability of adaptive neural network come Online Estimation timevariant nonlinear function
Wherein,It is the weight matrix of adaptive neural network,It is adaptive neural network
Basic function, l are the number of nodes of adaptive neural network hidden layer, ε_{j}It is the approximate error of adaptive neural network；
Wherein,It is adaptive neural network ideal weight matrixJth row realtime estimation value,It is
The basic function of adaptive neural network, l are the number of nodes of adaptive neural network hidden layer, ε_{j}It is forcing for adaptive neural network
Nearly error；
Therefore, the control based on model reference adaptive ANN Control inputs u_{j}, adaptive neural network weight is more
New matrix update ruleWith interference compensation control amount r_{j}Expression formula be respectively
r_{j}=δ_{j}sgn(e^{T}PB_{*j})
Wherein, e is the state variable x of current quantity of state x and expectational model^{mm}Difference, matrix P is the positive definite of state equation
Symmetric solution,
PA^{mm}+A^{mmT}P=Q
WhereinIt is positive definite symmetrical matrix,It is adaptive law gain matrix, B_{*j}It is the jth of B matrix
The upper bound of column, adaptive neural network evaluated error and external disturbance is
δ_{j}>ε_{j}x+d_{j}/b_{j}
Wherein, ε_{j}It is the approximate error of adaptive neural network, d_{j}For the interference of system j frame；
(2) adaptive neural network is constructed
For the interference compensation control amount r of the upper bound of adaptive neural network evaluated error and external disturbance building_{j}To being
The influence of system control precision, building extended state observer is estimated and is inhibited to interference, realizes three axis under complex environment
Inertially stabilized platform highprecision control,
Definition status variable z_{1}=e, z_{2}=Δ=B ε x+Dd, then three axis inertially stabilized platform error state equations be
By z_{2}Error state equation is introduced, and
Wherein, s is the rate of change of Δ, and the extended state observer of building is
Wherein,WithRespectively state variable z_{1}And z_{2}Estimated value, w_{0}> 0 is design variable；
Pass through suitable Selecting All Parameters w_{0}It can makeWithTight tracking e and Δ within the limited time,
Revised interference compensation control amount r_{j}Expression formula be
Wherein,Residual error B ε x and system interference Dd and disturbance estimated value are estimated for system selfadaption neural network
z_{2}Residual error
Corresponding system, which inputs, is
Wherein,It is+ 3 components of jth.
(3) flight example
In flight course, according to the angle information of highprecision attitude measuring unit, three axis inertially stabilized platform frame systems
System is adjusted correspondingly, and guarantees that the remote sensing load optical axis is vertical over the ground, flight result such as Fig. 2, Fig. 3 and Fig. 4 institute of certain experiment
Show.
Three axis inertially stabilized platforms realize highprecision control, and the standard deviation of pitch channel is 0.0183 degree, and roll is logical
The standard deviation in road is 0.0157, and the standard deviation of azimuthal channel is 0.0214.
The present invention is based on three axis Inertiallystabilizeplatform platform control methods of model reference adaptive neural network overcome it is existing
Highprecision control of the three axis inertially stabilized platforms in the case where disturbing environment complexity may be implemented in the deficiency of control method more.
The content that description in the present invention is not described in detail belongs to the prior art well known to professional and technical personnel in the field.
Claims (3)
1. a kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network, it is characterised in that realize
Following steps:
(1) according to three axis inertially stabilized platform kinetic models, design reference model and feedback controller, feedback control ginseng is solved
NumberFor the uncertain caused feedback control parameters timevarying characteristics of three axis inertially stabilized platform kinetic parameters, structure
The feedback control parameters of adaptive neural network Online Estimation timevariant nonlinear function are built,
Wherein,It is the weight matrix of adaptive neural network,It is the base letter of adaptive neural network
Number, l is the number of nodes of adaptive neural network hidden layer,For the state variable of system, n=6 is the dimension of state variable
Number, ε_{j}It is the approximate error of adaptive neural network jth row, when j=1 indicates roll frame, j=2 pitching frame, expression side when j=3
Position frame makes three axis inertially stabilized platforms control precision approach expectational model control precision；
(2) the interference compensation control for the upper bound building interfered for adaptive neural network evaluated error and three axis inertially stabilized platforms
Influence of the amount processed to system control precision constructs extended state observer and interference is estimated and inhibited,
Wherein,WithRespectively state variable z_{1}And z_{2}Estimated value, z_{1}=e, z_{2}=Δ=B ε x+Dd, e is current quantity of state x
With the state variable x of reference model^{mm}Difference,It is adaptive neural network weight square realtime estimation value,For frame
The coefficient matrix of frame input, m=3 are the dimensions of system input, and ε is the approximate error of adaptive neural network, and d is system framework
Interference, For the state equation matrix of reference model,WithIt is artificial based on POLE PLACEMENT USING principle according to dynamic performance requirement
The matrix of design, r are interference compensation control amount, w_{0}> 0 is design variable, realizes three axis inertially stabilized platforms under complex environment
Highprecision control.
2. the three axis inertially stabilized platform controlling parties according to claim 1 based on model reference adaptive neural network
Method, it is characterised in that: the step (1) is for the uncertain caused feedback control of three axis inertially stabilized platform kinetic parameters
Parameter time varying characteristic processed constructs the feedback control parameters of adaptive neural network Online Estimation timevariant nonlinear functionThree
The kinetic model of axis inertially stabilized platform is expressed as
Wherein, x=[θ_{j} ω_{j}]^{T},H=0_{3×3}, F=(f_{jk}), j, k=1,2,3,
U=[u_{1} u_{2} u_{3}]^{T}, g=[g_{1} g_{2} g_{3}]^{T}, d=[d_{1} d_{2} d_{3}]^{T},
Wherein, θ_{j}For corresponding j frame angle, ω_{j}For corresponding j frame angular speed,For the coefficient square of state variable
Battle array, u_{j}For the input of corresponding j frame voltage, g is the estimated value of the gimbal moment of intertia, d_{j}For the interference of system j frame, by system
It interferes, the perturbation value △ g of the gimbal moment of intertia_{j}With the control input perturbation △ b as caused by measurement noise_{j}It constitutes, b_{j}It is defeated for frame
The ideal control coefrficient of the corresponding j column of the coefficient matrix B entered, F are the corresponding perfect condition variation coefficient square with timevarying characteristics
Battle array；
Design reference model
WhereinIt is the control input of desired reference model；
Design of feedback controller
Wherein r_{j}It is the control amount for the corresponding frame interfering d for compensation and designing, meets It is
Feedback control parameters；
Then system
Feedback control parameters
Due to h_{jk},f_{jk}, k, j=1,2,3 be timevariant nonlinear function,It is also timevariant nonlinear function, therefore, utilization is adaptive
The infinite extensive approximation capability of neural network is answered to carry out Online Estimation timevariant nonlinear function
Control based on model reference adaptive ANN Control inputs u_{j}, adaptive neural network more new lawIt is mended with interference
Repay control amount r_{j}Expression formula be respectively
r_{j}=δ_{j}sgn(e^{T}PB_{*j})
Wherein, B_{*j}For the jth column of the coefficient matrix B of frame input, matrix P is the positive definite symmetric solution of state equation,
PA^{mm}+A^{mmT}P=Q
WhereinIt is positive definite symmetrical matrix,It is adaptive law gain matrix, adaptive neural network estimation misses
The upper bound of difference and external disturbance is
δ_{j}>ε_{j}x+d_{j}/b_{j}
Wherein, ε_{j}It is the approximate error of adaptive neural network.
3. the three axis inertially stabilized platform controlling parties according to claim 1 based on model reference adaptive neural network
Method, it is characterised in that: step (2) the building extended state observer is estimated and inhibited to interference, realizes complex environment
Under three axis inertially stabilized platform highprecision controls, three axis inertially stabilized platform error state equations are
By z_{2}Error state equation is introduced, andWherein, s is the rate of change of Δ, and the extended state observer of building is
Revised interference compensation control amount r_{j}Expression formula be
Wherein, b_{j}For the ideal control coefrficient of the coefficient matrix B corresponding j column of frame input,It is+ 3 components of jth,Residual error B ε x and system interference Dd and disturbance estimated value z are estimated for system selfadaption neural network_{2}Residual errord_{j}For the interference of system j frame, by the perturbation value △ g of system interference, the gimbal moment of intertia_{j}With
Input perturbation △ b is controlled as caused by measurement noise_{j}It constitutes, ε_{j}It is the approximate error of adaptive neural network, when j=1 indicates horizontal
Frame is rolled, j=2 pitching frame indicates orientation frame, B when j=3_{*j}For the jth column of the coefficient matrix B of frame input, matrix P is state side
The positive definite symmetric solution of journey,
PA^{mm}+A^{mmT}P=Q
WhereinIt is positive definite symmetrical matrix；
Corresponding system, which inputs, is
Wherein,It is the corresponding j column control input of desired reference model, g_{j}For the estimated value of the corresponding j gimbal moment of intertia.
Priority Applications (1)
Application Number  Priority Date  Filing Date  Title 

CN201610645413.4A CN106200383B (en)  20160808  20160808  A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network 
Applications Claiming Priority (1)
Application Number  Priority Date  Filing Date  Title 

CN201610645413.4A CN106200383B (en)  20160808  20160808  A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network 
Publications (2)
Publication Number  Publication Date 

CN106200383A CN106200383A (en)  20161207 
CN106200383B true CN106200383B (en)  20191018 
Family
ID=57514611
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

CN201610645413.4A Active CN106200383B (en)  20160808  20160808  A kind of three axis Inertiallystabilizeplatform platform control methods based on model reference adaptive neural network 
Country Status (1)
Country  Link 

CN (1)  CN106200383B (en) 
Families Citing this family (2)
Publication number  Priority date  Publication date  Assignee  Title 

CN107894713B (en) *  20171020  20201106  东南大学  Highprecision control method for twoaxis inertial stabilization platform without coding sensing 
CN108469269A (en) *  20180119  20180831  天津大学  A kind of resonance point test system of wideband inertial reference stabilized platform 
Citations (6)
Publication number  Priority date  Publication date  Assignee  Title 

CN1790197A (en) *  20051222  20060621  安徽工业大学  Simple method for neural network decoupling of multivariable system based on model reference adaptive control 
CN102053628A (en) *  20091027  20110511  北京航空航天大学  Neural networkbased servo control system and method 
CN102298315A (en) *  20110621  20111228  河海大学常州校区  Adaptive control system based on radial basis function (RBF) neural network sliding mode control for microelectromechanical system (MEMS) gyroscope 
CN102437816A (en) *  20111025  20120502  武汉鑫通科创科技发展有限公司  Adaptive motor motion control apparatus based on neural network 
CN102508503A (en) *  20111101  20120620  北京航空航天大学  Compensation method based on generalized inner module for eccentric torque of threeshaft inertially stabilized platform 
CN105785762A (en) *  20160317  20160720  北京航空航天大学  Biaxis inertiallystabilized platform highprecision control method based on selfadaptive backstepping sliding mode 

2016
 20160808 CN CN201610645413.4A patent/CN106200383B/en active Active
Patent Citations (6)
Publication number  Priority date  Publication date  Assignee  Title 

CN1790197A (en) *  20051222  20060621  安徽工业大学  Simple method for neural network decoupling of multivariable system based on model reference adaptive control 
CN102053628A (en) *  20091027  20110511  北京航空航天大学  Neural networkbased servo control system and method 
CN102298315A (en) *  20110621  20111228  河海大学常州校区  Adaptive control system based on radial basis function (RBF) neural network sliding mode control for microelectromechanical system (MEMS) gyroscope 
CN102437816A (en) *  20111025  20120502  武汉鑫通科创科技发展有限公司  Adaptive motor motion control apparatus based on neural network 
CN102508503A (en) *  20111101  20120620  北京航空航天大学  Compensation method based on generalized inner module for eccentric torque of threeshaft inertially stabilized platform 
CN105785762A (en) *  20160317  20160720  北京航空航天大学  Biaxis inertiallystabilized platform highprecision control method based on selfadaptive backstepping sliding mode 
NonPatent Citations (3)
Title 

A compound control method based on the adaptive neural network and sliding mode control for inertial stable platform;YingZou，等;《Neurocomputing》;20150107(第155期);第286294页 * 
三轴式无人旋翼飞行器及自适应飞行控制系统设计;夏青元,等;《航空学报》;20130325;第34卷(第3期);第495508页 * 
基于PMI的三轴惯性稳定平台干扰力矩补偿方法研究;钟麦英,等;《仪器仪表学报》;20140430;第35卷(第4期);第782787页 * 
Also Published As
Publication number  Publication date 

CN106200383A (en)  20161207 
Similar Documents
Publication  Publication Date  Title 

Xu et al.  Neural network based dynamic surface control of hypersonic flight dynamics using smallgain theorem  
CN105353615B (en)  A kind of Active Faulttolerant Control Method of the quadrotor based on sliding mode observer  
Hamel et al.  Evolution of flight vehicle system identification  
Liu et al.  Tracking control of smallscale helicopters using explicit nonlinear MPC augmented with disturbance observers  
CN102862686B (en)  Opticalintegral slidingmode attitude control method of reentry vehicle and controller  
CN106325291B (en)  Sliding mode control law and ESO (electronic stability program) based fourrotor aircraft attitude control method and system  
CN105911866B (en)  The finite time fullorder sliding mode control method of quadrotor unmanned vehicle  
Ren et al.  A multiposition selfcalibration method for dualaxis rotational inertial navigation system  
CN103913181B (en)  A kind of airborne distributed POS Transfer Alignments based on parameter identification  
CN103616818B (en)  The neural overall fast terminal slidingmode control of adaptive fuzzy of gyroscope  
CN105629732B (en)  A kind of spacecraft attitude output Tracking Feedback Control method for considering Control constraints  
CN102252673B (en)  Correction method for ontrack aberration of star sensor  
CN105094138A (en)  Lowaltitude autonomous navigation system for rotarywing unmanned plane  
CN104281056B (en)  The gyroscope Robust Adaptive Control method learnt based on the neutral net upper bound  
CN103336435B (en)  Gyroscope is based on the method for adaptive fuzzy sliding mode control of Attitude rate estimator  
Luis et al.  Design of a trajectory tracking controller for a nanoquadcopter  
CN105759832A (en)  Fourrotor aircraft sliding mode variable structure control method based on inversion method  
CN102073755B (en)  Motion control simulation method for nearspace hypersonic aircraft  
CN105159305B (en)  A kind of quadrotor flight control method based on sliding moding structure  
CN103900559B (en)  A kind of highprecision attitude resolving system based on Interference Estimation  
CN102506892B (en)  Configuration method for information fusion of a plurality of optical flow sensors and inertial navigation device  
CN103344243B (en)  A kind of aerial remote sensing inertialstabilized platform friction parameter discrimination method  
CN103365296A (en)  Nonlinear output feedback flight control method for quadrotor unmanned aerial vehicle  
CN105806365B (en)  Rapid Alignment Technology between a kind of vehiclemounted inertial navigation based on Active Disturbance Rejection Control is advanced  
CN107608367B (en)  Multivariable interference compensation quadrotor unmanned aerial vehicle trajectory and attitude cooperative control method 
Legal Events
Date  Code  Title  Description 

PB01  Publication  
C06  Publication  
SE01  Entry into force of request for substantive examination  
C10  Entry into substantive examination  
GR01  Patent grant  
GR01  Patent grant 