Loading [MathJax]/jax/element/mml/optable/BasicLatin.js
Xiao-Yi Wang and Guang-Ren Duan. A Direct Parametric Approach to Spacecraft Attitude Tracking Control. International Journal of Automation and Computing, vol. 14, no. 5, pp. 626-636, 2017. DOI: 10.1007/s11633-017-1089-4
Citation: Xiao-Yi Wang and Guang-Ren Duan. A Direct Parametric Approach to Spacecraft Attitude Tracking Control. International Journal of Automation and Computing, vol. 14, no. 5, pp. 626-636, 2017. DOI: 10.1007/s11633-017-1089-4

A Direct Parametric Approach to Spacecraft Attitude Tracking Control

Funds: 

National Natural Science Foundation of China 61321062

More Information
  • Author Bio:

    Guang-Ren Duan: Guan-Ren Duan received the B. Sc. degree in applied mathematics from Yanshan University, China in 1983, the M. Sc. degree in control systems theory from Harbin Engineering University, China in 1986, and received the Ph. D. degree in control systems theory from Harbin Institute of Technology, China in 1989. From 1989 to 1991, he was a post-doctoral researcher at Harbin Institute of Technology, where he became a professor of control systems theory in 1991. He visited the University of Hull, UK, and the University of Sheffield, UK from December 1996 to October 1998, and worked at the Queens University of Belfast, UK from October 1998 to October 2002. Since August 2000, he has been elected specially employed professor at Harbin Institute of Technology sponsored by the Cheung Kong Scholars Program of the Chinese government. He is currently the director of the Centre for Control Systems and Guidance Technology at Harbin Institute of Technology. He is the author and co-author of over 300 publications. He is a chartered engineer in the UK, a senior member of IEEE and a fellow of IEE.
         His research interests include robust control, eigenstructure assignment, descriptor systems, missile autopilot control and magnetic bearing control.
         E-mail:g.r.duan@hit.edu.cn

  • Corresponding author:

    Xiao-Yi Wang received the B. Sc. degree from Department of Control Science and Engineering, Harbin Institute of Technology, China in 2014. She received the M. Sc. degree in Centre for Control Theory and Guidance Technology, Harbin Institute of technology, China in 2017.
         Her research interests include nonlinear robust control and spacecraft attitude control.
         E-mail:xiaoyiwang51@outlook.com (Corresponding author)
         ORCID iD:0000-0002-8511-1273

  • Received Date: August 18, 2016
  • Accepted Date: March 28, 2017
  • Available Online: July 02, 2017
  • Published Date: July 02, 2017
  • Through the direct parameter approach, a solution for spacecraft attitude tracking is presented. First of all, the spacecraft attitude tracking control model is built up by the error equation of the second-order nonlinear quaternion-based attitude system. Based on the control model, a suitable controller is designed by the direct parameter approach. Compared with other control strategies, the direct parameter approach can offer all degrees of freedom for the controller to satisfy the requirements for system properties and turn the original nonlinear system into closed-loop linear system. Furthermore, this paper optimizes the controller according to the robustness, limitation of controller output and closed-loop eigenvalue sensitivity. Putting the controller into the original system, the state response of the closed-loop system and the output of controller are plotted in Matlab to verify the availability and robustness of the controller. Therefore, the controlled spacecraft can achieve the goal of tracking on the mobile target with the external disturbance torque.
  • Spacecraft attitude control has been a hot problem for several years as a significant part of spacecraft navigation control. Many different methods are applied on the problem of spacecraft and aircraft attitude control [1-20]. For instance, Parlos and Sunkel [4] used linear quadratic Gaussian (LQG) control to solve a type of large spacecraft attitude angular maneuver. Bang et al. [5] designed a new sliding model control method which can deal with flexible spacecraft attitude maneuver problem. Singh and Zhang [6] utilized adaptive output feedback control method to track the flexible spacecraft attitude. Additionally, Guan [1, 2, 21] raised a direct parametric approach to stabilize the nonlinear second-order spacecraft attitude model, and the controller can turn the original model into a stable linear constant system.

    According to the research background of spacecraft attitude control, spacecraft attitude tracking needs further research and has more profound applications. Compared with Euler representation, quaternion representation can describe large-angle attitude tracking without singular points which should appear when the changing of Euler angle is bigger than 90°. Moreover, this paper raises a fully-actuated second-order system based on the spacecraft attitude dynamical and kinematical differential equations. Although most of methods deal with the spacecraft attitude control problem by the first-order models, this paper takes advantage of the fully-actuated second-order model and makes the process of controller design easier and clearer than methods based on the first-order models. Finally, this paper builds up a control model for the spacecraft attitude tracking through the error equation between controlled object and the mobile target. According to the direct parametric approach provided by Duan in [1, 2, 22-24], a controller is designed to stabilize the control model, which means the controlled spacecraft successfully tracks the target. To design the controller, the first step is to compensate the nonlinear term ξ of the non-approximated control model by controller uc then obtain the quasi-linear system. The next step is to convert the stabilization problem of the quasi-linear system into the problem of finding the solution for relative Sylvester equation, and yield the expression of the state feedback controller uf . Moreover, the controller uf can be optimized by the free parameters F2n×2n and Z2n×n . All the elements in F and Z can be arbitrary, thus the direct parametric approach could provide all degrees of freedom of the prospective controller. With the useful advantage of the approach, the controller can be easily designed to satisfy the requirement of performance of closed-loop system. By the adopted way, this method can turn the original highly nonlinear model into a linear closed-loop constant system. Finally, the closed-loop system is simulated in Matlab to verify the availability and robustness of the controller when there is certain external disturbance torque applied on the control model.

    In this section, a second-order spacecraft attitude control model is built by attitude dynamical equations and kinematical equations. Additionally, that model is described by quaternion and without any approximation. The problem of quaternion-based spacecraft attitude tracking is raised based on the attitude control model.

    According to the attitude dynamics and kinematics of a rigid spacecraft in the inertial frame, the error attitude dynamics of a rigid spacecraft relative to a mobile tracking target in the inertial frame is given by

    J˙ωe=ω×Jω+J(ω×C(qe)ωrC(qe)˙ωr)+u+d

    (1)

    where

    J=diag(Jx,Jy,Jz)

    (2)

    is the rotating inertia matrix of the spacecraft, ω is the angular rate vector, qe is the error attitude quaternion between controlled spacecraft and mobile target, ωe is the error angular rate vector relative to the mobile tracking target, ωr is the angular rate vector for target, u is the input control torque vector, and d is the disturbance torque vector. Besides, ω , ωe , ωr , u and d belong to R1×3 .

    In addition, C(qe) is the transition matrix and given by [19, 20]

    C(qe)=[C1(qe)   C2(qe)   C3(qe)]

    (3)

    where

    C1(qe)=[q2e1q2e2q2e3+q2e02(qe1qe2qe3qe0)2(qe1qe3+qe2qe0)] 

    (4)

    C2(qe)=[2(qe1qe2+qe3qe0)q2e1+q2e2q2e3+q2e02(qe2qe3qe1qe0)]

    (5)

    C3(qe)=[2(qe1qe3qe2qe0)2(qe2qe3+qe1qe0)q2e1q2e2+q2e3+q2e0]

    (6)

    qe=[qe0qe1qe2qe3]T.

    (7)

    The variable ωe can be expressed by

    ωe=ωC(qe)ωr.

    (8)

    Then, the kinematics of a rigid spacecraft in the inertial frame based on error quaternion is given by

    [˙qe0˙qe1˙qe2˙qe3]=12[qe1qe2qe3qe0qe3qe2qe3qe0qe1qe2qe1qe0][ωexωeyωez] 

    (9)

    where ωe is the error angular rate vector relative to the mobile tracking target, and qe is the error quaternion relative to the target.

    When the system is stable, this means

    {lim

    (10)

    To stabilize the controlled system, a suitable controller {u} is given by

    {u} = f\left( {{{\omega _e}}, {q_{e0}}, {q_{e1}}, {q_{e2}}, {q_{e3}}, t} \right).

    (11)

    When the controlled system is stable, the spacecraft error attitude q_{e} will successfully achieve the stable equilibrium point [1, 0, 0, 0]T, which means that the controlled spacecraft successfully tracks on the mobile target.

    Generally, the error quaternion q_{e} has the constraint which is

    {q_{e0}}^2 + q_{e1}^2 + q_{e2}^2 + q_{e3}^2 = 1\

    (12)

    {e}={\left[{\begin{array}{*{20}{c}} {{e_1}} & {{e_2}} & {{e_3}} \end{array}} \right]^{\rm T}}={\left[{\begin{array}{*{20}{c}} {{q_{e1}}} & {{q_{e2}}} & {{q_{e3}}} \end{array}} \right]^{\rm T}}.\

    (13)

    Additionally, q_{e} _{0} can be denoted as

    {q_{e0}} = {e_0} = \sqrt {1 - \left( {e_1^2 + e_2^2 + e_3^2} \right)}.

    (14)

    With the expression of e , we get from (9) that

    \dot e = \frac{1}{2}T\left( e \right){\omega _e} = \frac{1}{2}\left[{\begin{array}{*{20}{c}} {{e_0}} & {-{e_3}} & {{e_2}}\\ {{e_3}} & {{e_0}} & {-{e_1}}\\ {-{e_2}} & {{e_1}} & {{e_0}} \end{array}} \right]{\omega _e}.

    (15)

    Taking derivative of (15) with respect to time t gives

    \ddot{ {e}}= \frac{1}{2}\frac{{{\rm d}{{T}}\left( {e} \right)}}{{{\rm d}t}}{{\omega _e}} + \frac{1}{2}{ {T}}\left( { {e}} \right){\dot {{\omega _e}}}.

    (16)

    The expression for {\omega_e} can be derived from (15) as

    \begin{array}{*{35}{l}} {{\omega }_{e}}= & 2\left[ -\dot{e}\left( e \right){{T}^{\text{T}}}\left( e \right) \right]{{\left[ -\frac{1}{2}{{e}^{\text{T}}}\omega \dot{e} \right]}^{\text{T}}}= \\ {} & 2\left( \frac{1}{2}e{{e}^{\text{T}}}{{\omega }_{e}}+{{T}^{\text{T}}}\left( e \right)\dot{e} \right)= \\ {} & e{{e}^{\text{T}}}{{\omega }_{e}}+2{{T}^{\text{T}}}\left( e \right)\dot{e}= \\ {} & 2{{\left( {{I}_{3}}-e{{e}^{\text{T}}} \right)}^{-1}}{{T}^{\text{T}}}\left( e \right)\dot{e}= \\ {} & 2\Phi \left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e} \\ \end{array}

    (17)

    where

    {\Phi} \left( {e} \right) = {\left( {{{I}_3} - {e{e^{\rm T}}}} \right)^{ - 1}} = \left[{\begin{array}{*{20}{c}} {{\Phi _1}\left( e \right)}\\ {{\Phi _2}\left( e \right)}\\ {{\Phi _3}\left( e \right)} \end{array}} \right].

    (18)

    Thus,

    \left\{ \begin{array}{*{35}{l}} {{\omega }_{ex}}=2{{\Phi }_{1}}\left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e} \\ {{\omega }_{ex}}=2{{\Phi }_{2}}\left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e} \\ {{\omega }_{ex}}=2{{\Phi }_{3}}\left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e}. \\ \end{array} \right.

    (19)

    The next step is to summarize the second-order spacecraft control model through (1) and (16). First of all, because of

    {\omega } = {\omega _e} - {C}\left( {e} \right){\omega _r}.

    (20)

    Equation (1) can be written as

    \begin{array}{*{35}{l}} J{{{\dot{\omega }}}_{e}}= & -\left( {{\omega }_{e}}-C\left( e \right){{\omega }_{r}} \right)\times J\left( {{\omega }_{e}}-C\left( e \right){{\omega }_{r}} \right)+ \\ {} & J\left( \left( {{\omega }_{e}}-C\left( e \right){{\omega }_{r}} \right)\times C\left( {{q}_{e}} \right){{\omega }_{r}}-C\left( {{q}_{e}} \right){{{\dot{\omega }}}_{r}} \right)+ \\ {} & u+d. \\ \end{array}

    (21)

    Then, the expression of \dot{\omega _e} can be derived by (21) as

    \begin{array}{*{35}{l}} {{{\dot{\omega }}}_{e}}= & {{J}^{-1}}\left[ {{\omega }_{e}}\times \left( J{{\omega }_{e}} \right) \right]- \\ {} & {{J}^{-1}}\left\{ {{\omega }_{e}}\times \left[ JC\left( e \right){{\omega }_{r}} \right] \right\}- \\ {} & {{J}^{-1}}\left\{ \left[ C\left( e \right){{\omega }_{r}} \right]\times \left[ J{{\omega }_{e}} \right] \right\}- \\ {} & {{J}^{-1}}\left\{ \left[ C\left( e \right){{\omega }_{r}} \right]\times \left[ JC\left( e \right){{\omega }_{r}} \right] \right\}+ \\ {} & \left[ {{\omega }_{e}}\times \left( C\left( e \right){{\omega }_{r}} \right) \right]-C\left( e \right){{{\dot{\omega }}}_{r}}+ \\ {} & {{J}^{-1}}\left( u+d \right). \\ \end{array}

    (22)

    Denote

    {\Gamma \left( {J, e, \dot e} \right)} = \left[{\begin{array}{*{20}{c}} 0 & 0 & {{\Gamma _c}}\\ {{\Gamma _a}} & 0 & 0\\ 0 & {{\Gamma _b}} & 0 \end{array}} \right]\

    (23)

    \left\{ \begin{array}{l} {\Gamma _a} = \dfrac{{{J_z} - {J_x}}}{{{J_y}}}{{\Phi _3}}\left( {e} \right){{T^{\rm T}}}\left( {e} \right)\dot {{e}}\\ {\Gamma _b} = \dfrac{{{J_x} - {J_y}}}{{{J_z}}}{{\Phi _1}}\left( {e} \right){{T^{\rm T}}}\left( {e} \right)\dot {{e}}\\ {\Gamma _c} = \dfrac{{{J_y} - {J_z}}}{{{J_x}}}{{\Phi _2}}\left( {e} \right){{T^{\rm T}}}\left( {e} \right)\dot {{e}} \end{array} \right.\nonumber\\[-5mm]

    (24)

    G = \left[{\begin{array}{*{20}{c}} {{G_1}}\\ {{G_2}}\\ {{G_3}} \end{array}} \right] = \left[{\begin{array}{*{20}{c}} {{{\left[{JC\left( e \right){\omega _r}} \right]}_x}}\\ {{{\left[{JC\left( e \right){\omega _r}} \right]}_y}}\\ {{{\left[{JC\left( e \right){\omega _r}} \right]}_z}} \end{array}} \right]\

    (25)

    and

    {H} = \left[{\begin{array}{*{20}{c}} {{H_1}}\\ {{H_2}}\\ {{H_3}} \end{array}} \right] = {\left[{\begin{array}{*{20}{c}} {{{\left[{C\left( e \right){\omega _r}} \right]}_x}}\\ {{{\left[{C\left( e \right){\omega _r}} \right]}_y}}\\ {{{\left[{C\left( e \right){\omega _r}} \right]}_z}} \end{array}} \right]}.

    (26)

    The expression for {\dot{\omega_e}} (22) can be expanded as

    \begin{array}{*{35}{l}} {{{\dot{\omega }}}_{e}}= & 4\Gamma \left( J,e,\dot{e} \right)\Phi \left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e}- \\ {} & {{J}^{-1}}\left[ \begin{matrix} {{G}_{3}}{{\Phi }_{2}}-{{G}_{2}}{{\Phi }_{3}} \\ {{G}_{1}}{{\Phi }_{3}}-{{G}_{3}}{{\Phi }_{1}} \\ {{G}_{2}}{{\Phi }_{1}}-{{G}_{1}}{{\Phi }_{2}} \\ \end{matrix} \right]2{{T}^{\text{T}}}\dot{e}- \\ {} & {{J}^{\text{-}1}}\left[ \begin{matrix} {{H}_{2}}{{J}_{z}}{{\Phi }_{3}}-{{H}_{3}}{{J}_{y}}{{\Phi }_{2}} \\ {{H}_{3}}{{J}_{x}}{{\Phi }_{1}}-{{H}_{1}}{{J}_{z}}{{\Phi }_{3}} \\ {{H}_{1}}{{J}_{y}}{{\Phi }_{2}}-{{H}_{2}}{{J}_{x}}{{\Phi }_{1}} \\ \end{matrix} \right]2{{T}^{\text{T}}}\dot{e}+ \\ {} & \left[ \begin{matrix} {{H}_{2}}{{\Phi }_{3}}-{{H}_{3}}{{\Phi }_{2}} \\ {{H}_{3}}{{\Phi }_{1}}-{{H}_{1}}{{\Phi }_{3}} \\ {{H}_{1}}{{\Phi }_{2}}-{{H}_{2}}{{\Phi }_{1}} \\ \end{matrix} \right]2{{T}^{\text{T}}}\dot{e}- \\ {} & {{J}^{-1}}\left[ H\times \left( JH \right) \right]-C\left( e \right){{{\dot{\omega }}}_{r}}+ \\ {} & {{J}^{-1}}\left( u+d \right). \\ \end{array}

    (27)

    Substituting (17), (22) and (26) into (16) gives

    \begin{array}{*{35}{l}} \ddot{e}= & \frac{1}{2}\frac{\text{d}T\left( e \right)}{\text{d}t}{{\omega }_{e}}+\frac{1}{2}T\left( e \right){{{\dot{\omega }}}_{e}}= \\ {} & \frac{1}{2}\frac{\text{d}T\left( e \right)}{\text{d}t}\left[ 2\Phi \left( e \right){{T}^{\text{T}}}\left( e \right)\dot{e} \right]= \\ {} & \left( \begin{array}{*{35}{l}} \frac{\text{d}T}{\text{d}t}\Phi \left( e \right)+2T\Gamma \left( J,e,\dot{e} \right)\Phi \left( e \right)+ \\ {{J}^{-1}}T\left[ \begin{matrix} {{G}_{3}}{{\Phi }_{2}}-{{G}_{2}}{{\Phi }_{3}} \\ {{G}_{1}}{{\Phi }_{3}}-{{G}_{3}}{{\Phi }_{1}} \\ {{G}_{2}}{{\Phi }_{1}}-{{G}_{1}}{{\Phi }_{2}} \\ \end{matrix} \right]+ \\ {{J}^{\text{-}1}}T\left[ \begin{matrix} {{H}_{2}}{{J}_{z}}{{\Phi }_{3}}-{{H}_{3}}{{J}_{y}}{{\Phi }_{2}} \\ {{H}_{3}}{{J}_{x}}{{\Phi }_{1}}-{{H}_{1}}{{J}_{z}}{{\Phi }_{3}} \\ {{H}_{1}}{{J}_{y}}{{\Phi }_{2}}-{{H}_{2}}{{J}_{x}}{{\Phi }_{1}} \\ \end{matrix} \right]- \\ T\left[ \begin{matrix} {{H}_{2}}{{\Phi }_{3}}-{{H}_{3}}{{\Phi }_{2}} \\ {{H}_{3}}{{\Phi }_{1}}-{{H}_{1}}{{\Phi }_{3}} \\ {{H}_{1}}{{\Phi }_{2}}-{{H}_{2}}{{\Phi }_{1}} \\ \end{matrix} \right] \\ \end{array} \right){{T}^{\text{T}}}\left( e \right)\dot{e}- \\ {} & \frac{1}{2}T\left\{ {{J}^{-1}}\left[ H\times \left( JH \right) \right]-C\left( e \right){{{\dot{\omega }}}_{r}} \right\}+ \\ {} & \frac{1}{2}T{{J}^{-1}}\left( u+d \right). \\ \end{array}

    (28)

    In addition, the second-order spacecraft attitude tracking control model (28) can be written in the second-order form as

    \left\{ \begin{array}{l} {{A_2}\left( {\theta, e, \dot e} \right)\ddot e} + {{A_1}\left( {\theta, e, \dot e} \right)\dot e} +\\ {{A_0}\left( {\theta, e, \dot e} \right)e} + {\xi} \left( {{\theta}, {e}, {\dot{ e}}, t} \right) \end{array} \right\} = {B\left( {\theta, e, \dot e} \right)u}

    (29)

    where

    {{A_2}\left( {\theta, e, \dot e} \right)} = {I_3}

    (30)

    \begin{align} & {{A}_{1}}\left( \theta ,e,\dot{e} \right)= \\ & \quad \quad \quad \left( \begin{array}{*{35}{l}} \frac{\text{d}T}{\text{d}t}\Phi \left( e \right)+2T\Gamma \left( J,e,\dot{e} \right)\Phi \left( e \right)+ \\ {{J}^{-1}}T\left[ \begin{matrix} {{G}_{3}}{{\Phi }_{2}}-{{G}_{2}}{{\Phi }_{3}} \\ {{G}_{1}}{{\Phi }_{3}}-{{G}_{3}}{{\Phi }_{1}} \\ {{G}_{2}}{{\Phi }_{1}}-{{G}_{1}}{{\Phi }_{2}} \\ \end{matrix} \right]+ \\ {{J}^{\text{-}1}}T\left[ \begin{matrix} {{H}_{2}}{{J}_{z}}{{\Phi }_{3}}-{{H}_{3}}{{J}_{y}}{{\Phi }_{2}} \\ {{H}_{3}}{{J}_{x}}{{\Phi }_{1}}-{{H}_{1}}{{J}_{z}}{{\Phi }_{3}} \\ {{H}_{1}}{{J}_{y}}{{\Phi }_{2}}-{{H}_{2}}{{J}_{x}}{{\Phi }_{1}} \\ \end{matrix} \right]- \\ T\left[ \begin{matrix} {{H}_{2}}{{\Phi }_{3}}-{{H}_{3}}{{\Phi }_{2}} \\ {{H}_{3}}{{\Phi }_{1}}-{{H}_{1}}{{\Phi }_{3}} \\ {{H}_{1}}{{\Phi }_{2}}-{{H}_{2}}{{\Phi }_{1}} \\ \end{matrix} \right] \\ \end{array} \right)\times {{T}^{\text{T}}}\left( e \right)\dot{e} \\ \end{align}

    (31)

    {{A_0}\left( {\theta, e, \dot e} \right)} = {0_3}

    (32)

    {B\left( {\theta, e, \dot e} \right)} = \frac{1}{2}{T\left( e \right){J^{ - 1}}}

    (33)

    {\xi} \left( {{\theta}, {e}, {\dot e}, t} \right) = \frac{1}{2}{T}\left\{ \begin{array}{l} {{J^{ - 1}}\left[{H \times \left( {JH} \right)} \right]}- \\ {C\left( e \right){{\dot \omega }_r} - {J^{ - 1}}d} \end{array} \right\}

    (34)

    where A_{0} , A_{1} , A_{2} are the coefficient matrices, and {B} is the input matrix, {\xi} is the nonlinear term, {u} is the controller, {d} is the external disturbance.

    When the second-order system is stable, which means

    \left\{ {\begin{array}{*{20}{l}} {\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {e_1} = 0,\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {e_2} = 0,\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {e_3} = 0}\\ {\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {{\dot e}_1} = 0,\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {{\dot e}_2} = 0,\mathop {\lim }\limits_{t \to \infty } {\mkern 1mu} {{\dot e}_3} = 0.} \end{array}} \right.

    (35)

    To stabilize the controlled system (28), a suitable controller {u} is given by

    u = f\left( {{e_1}, {e_2}, {e_3}, t} \right).

    (36)

    When the controlled system is stable, the spacecraft error attitude {e} will successfully achieve the stable equilibrium point [0, 0, 0]T, this means that the controlled spacecraft successfully tracks the mobile target.

    To use the direct parametric approach provided by Duan in [1] to design controller for the control model, there are three assumptions needed to satisfy.

    Assumption 1. The values of the system parameter \theta =\theta (t)\in \Omega \subset {{\bf{R}}^{l}} .

    Define the parameter \theta as

    {\theta} \left( t \right) = \left[{\begin{array}{*{20}{c}} {{J_x}\left( t \right)} & {{J_y}\left( t \right)} & {{J_z}\left( t \right)} \end{array}} \right]

    (37)

    where {{\text{J}}_{i}}(i=x,y,z) is the rotational inertia of one axis.

    Particularly, the case of space capturing is considered. It is supposed that the two spacecrafts are joined together at the time t1, then they departed from each other at time t2. The rotational inertia can be defined as

    {{J}_{i}}\left( t \right)=\left\{ \begin{array}{*{35}{l}} \begin{matrix} {{J}_{i1}}, & t<{{t}_{1}} \\ \end{matrix} \\ \begin{matrix} {{J}_{i2}}, & {{t}_{1}}\le t<{{t}_{2}} \\ \end{matrix} \\ \begin{matrix} {{J}_{i3}}, & t\le {{t}_{2}} \\ \end{matrix} \\ \end{array} \right.i=x,y,z

    (38)

    where J_{xi} , J_{yi} and J_{zi} (i = 1, 2) are constant positive scalars. Thus, parameter \theta (t) satisfies the Assumption 1.

    Assumption 2. det( {{A} _2 }) \neq 0, \forall {{e}}, \space{{\dot{e} }} and {{ \theta }} ({t}) \in \space{\Omega}.

    According to (29), det ( {A} _2 ) = 1. So the system should satisfy Assumption 2.

    Assumption 3. {B} is uniformly bounded and det ({{B}}) \neq 0, \forall {{e}} , \space{{ \dot{e} }} and {{ \theta }} ({t}) \in \space{ \Omega}.

    According to

    {B} = \frac{1}{2}{T\left( e \right){J^{ - 1}}}

    (39)

    \det \left( {B} \right) = \frac{1}{{8{J_x}{J_y}{J_z}}}{e_0}.

    (40)

    Assumption 3 is equivalent to {{e}_{0}}\ne 0 , because it has

    {e_0} = \sqrt {1 - \left( {e_1^2 + e_2^2 + e_3^2} \right)}

    (41)

    where {{e}_{0}}\ne 0 is equal to

    e_1^2 + e_2^2 + e_3^2 < 1.

    (42)

    Although this condition cannot be satisfied all the time, it can be met when the initial values of {e} are chosen carefully. When this condition is met, Assumption 3 is satisfied.

    A controller {u} is designed for the spacecraft attitude tracking model (29), which is made up of two parts as

    {u} = {u_c} + {u_f}

    (43)

    where u _c is the compensating controller for the term \xi , and its expression is

    \quad {u_c} = {\left( {\frac{1}{2}{T}\left( {e} \right){{{J}}^{ - 1}}} \right)^{ - 1}}\frac{1}{2}{T}\left\{ \begin{array}{l} {{J}^{ - 1}}\left[{H \times \left( {JH} \right)} \right] +\\ {C}\left( {e} \right){{\dot \omega }_r} - {{J^{ - 1}}d} \end{array} \right\}.\quad

    (44)

    In addition, {u} _f is the state feedback controller for the compensated control system and given by

    {u_f} = {K_0}\left( {\theta, e, \dot e} \right){e} + {K_1}\left( {\theta, e, \dot e} \right)\dot {{e}} + {v}

    (45)

    where {K} _0 and {K}_1 \in {\bf R}^{3\times 3} are the continuous state feedback gains, and {v} is an external signal.

    Then, a parametric controller {u} _f is going to be designed by a direct parametric approach.

    Define

    {F} = \left\{ \begin{array}{l} \left. {F} \right|{F} \in {{\bf R}^{2n \times 2n}}, {\rm and}~\exists {Z} \in {{\bf R}^{2n \times 2n}}\\ {\rm s.t.}~~\det \left[ {\begin{array}{*{20}{c}} {Z}\\ {ZF} \end{array}} \right] \ne 0 \end{array} \right\}

    (46)

    where {F} and {Z} are parametric matrices, and n is the dimension of the second-order system (29).

    Step 1. Find the Sylvester equation for the control model (29).

    Denote

    {{V^{ - 1}}{A_c}\left( {\theta, e, \dot e} \right)V} = {F}

    (47)

    {{V}_{2n \times 2n}} = \left[{\begin{array}{*{20}{c}} {{{\left( {{V_0}} \right)}_{n \times 2n}}}\\ {{{\left( {{V_1}} \right)}_{n \times 2n}}} \end{array}} \right] = \left[{\begin{array}{*{20}{c}} {{V_0}}\\ {{V_0}F} \end{array}} \right]

    (48)

    and

    \det \left( {V} \right) \ne 0

    (49)

    where {A} _c is the closed-loop system matrix, and {{V}} \in \textbf{R}^{2n\times2n} is a parametric matrix.

    Through the direct parametric approach, the relative Sylvester equation for the system (29) can be given by

    \left\{ \begin{array}{l} {{A_2}\left( {\theta, e, \dot e} \right){V_0}{F^2}} +\\ {{A_1}\left( {\theta, e, \dot e} \right){V_0}F} +\\ {{A_0}\left( {\theta, e, \dot e} \right){V_0}} \end{array} \right\} = {B\left( {\theta, e, \dot e} \right)W}

    (50)

    where {{W}} \in \textbf{R}^{n\times 2n} is the solution for (50).

    Step 2. Solve the Sylvester equation and gain the expression for feedback controller {{u}}_f .

    It is suggested by the direct parametric approach that the solution for the Sylvester equation is

    \left\{ \begin{array}{*{35}{l}} V=\left[ \begin{array}{*{35}{l}} {{N}_{0}}Z \\ {{N}_{0}}ZF \\ \end{array} \right]=\left[ \begin{matrix} Z \\ ZF \\ \end{matrix} \right] \\ W={{B}^{-1}}\left( {{A}_{2}}Z{{F}^{2}}+{{A}_{1}}ZF+{{A}_{0}}Z \right) \\ \end{array} \right.

    (51)

    and {V} satisfies the condition which is det ({{V}}) \neq 0, then the controller {{u} _f } can be written as

    {u_f} = {{K_0}\left( {\theta, e, \dot e} \right)e} + {{K_1}\left( {\theta, e, \dot e} \right)\dot e} + {v}

    (52)

    where

    {K_F} = \left[{\begin{array}{*{20}{c}} {{K_0}} & {{K_1}} \end{array}} \right] = {W}{{V}^{ - 1}}.

    (53)

    Denote

    {X} = \left[{\begin{array}{*{20}{c}} {e}\\ {\dot e} \end{array}} \right].

    (54)

    Then, the closed-loop system can be converted into the first-order form:

    {\dot X} = {{A_c}X} + {{B_c}v}

    (55)

    where

    {A_c} = \left[{\begin{array}{*{20}{c}} {0} & {{I_n}}\\ {{-{A_2}^{-1}A_0^c}} & {{-{A_2}^{ - 1}A_1^c}} \end{array}} \right]

    (56)

    {B_c} = \left[{\begin{array}{*{20}{c}} {0}\\ {{{A_2}^{-1}\left( {\theta, e, \dot e} \right)B\left( {\theta, e, \dot e} \right)}} \end{array}} \right]

    (57)

    and

    \qquad \left\{ \begin{array}{l} {A_0^c\left( {\theta, e, \dot e} \right)} = {{A_0}\left( {\theta, e, \dot e} \right) - B\left( {\theta, e, \dot e} \right){K_0}\left( {\theta, e, \dot e} \right)}\\ {A_1^c\left( {\theta, e, \dot e} \right)} ={ {A_1}\left( {\theta, e, \dot e} \right) - B\left( {\theta, e, \dot e} \right){K_1}\left( {\theta, e, \dot e} \right)}. \end{array}\qquad \right.

    (58)

    Step 3. Optimize the controller {{u} _f } by the optimizing index {{J}} _{opt} .

    To maintain the robustness of performance and robustness of stability for a controlled system with uncertain parameters, the beneficial method is to minimize the closed-loop eigenvalues sensitivities given by [21]. As the closed-loop system should be a constant stable linear system, the optimizing index could be chosen as

    \begin{matrix} {{J}_{opt}}={{J}_{opt}}\left( F,Z \right)=\left\| V \right\|\left\| {{V}^{-1}} \right\|= \\ \qquad \left\| \left[ \begin{matrix} Z \\ ZF \\ \end{matrix} \right] \right\|\left\| {{\left[ \begin{matrix} Z \\ ZF \\ \end{matrix} \right]}^{-1}} \right\|. \\ \end{matrix}

    (59)

    Furthermore, the constrained nonlinear programming problem can be described as

    \begin{array}{*{35}{l}} \min {{J}_{opt}}\left( F,Z \right) \\ \rm{s}.\rm{t}.\left\{ \begin{array}{*{35}{l}} \det \left( \begin{matrix} Z \\ ZF \\ \end{matrix} \right)\ne 0,~Z\in {{\bf{R}}^{n\times 2n}} \\ {{\lambda }_{i}}\in {{C}^{-}},~i=1,2,\cdots ,2n. \\ \end{array} \right. \\ \end{array}

    (60)

    Besides, to simplify the optimizing process, the parameter F can be considered as

    {F} ={\rm diag} \left( {{\lambda _1},{\lambda _2}, \cdots ,{\lambda _{2n}}} \right)

    (61)

    where \lambda_{i} is the eigenvalue of the closed-loop system.

    Step 4. Simulation.

    Put the controller u into the control model, then utilize Matlab to acquire the response for closed-loop system. Finally, analyze the performance of the closed-loop system and make the conclusion.

    A set of practical spacecraft attitude system control data example is provided for the simulation [3].

    The rotational inertia of the spacecraft is

    {J} = \left[{\begin{array}{*{20}{c}} {{J_x}} & 0 & 0\\ 0 & {{J_y}} & 0\\ 0 & 0 & {{J_z}} \end{array}} \right] = \left[{\begin{array}{*{20}{c}} {18} & 0 & 0\\ 0 & {21} & 0\\ 0 & 0 & {24} \end{array}} \right]{\rm kg\cdot{m^2}}

    The disturbance torque d is

    {d} = 1.0 \times {10^{ - 3}}\left[{\begin{array}{*{20}{c}} {\cos \left( {0.01t} \right)-0.3}\\ {0.3\cos (0.02t) + 0.6}\\ {0.5\sin (0.02t)} \end{array}} \right]{\rm{N}} \cdot {\rm{m}}.

    The initial attitude quaternion of controlled spacecraft {{q}} _d (0) is

    {q_d}\left( 0 \right) = {\left[{\begin{array}{*{20}{c}} 1 & 0 & 0 & 0 \end{array}} \right]^{\rm T}}.

    The initial attitude quaternion of target spacecraft {{q}} _r (0) is

    {q_r}\left( 0 \right) = {\left[{\begin{array}{*{20}{c}} {0.668\, 3} & {-0.554\, 6} & {0.399\, 9} & {0.293\, 1} \end{array}} \right]^{\rm T}}.

    As q_e (0) = q_{r}^{\ast} (0) \, \circ\, {{q}}_{d} (0), the initial error attitude quaternion {{q}} _{e} (0) is

    {q_e}\left( 0 \right) = {\left[{\begin{array}{*{20}{c}} {0.668\, 3} & {{\rm{0}}{\rm{.554\, 6}}} & {{\rm{-0}}{\rm{.399\, 9}}} & {{\rm{-0}}{\rm{.293\, 1}}} \end{array}} \right]^{\rm T}}.

    Thus, the initial state {e} (0) and \dot{e} (0) are

    {e}\left( 0 \right) = {\left[{\begin{array}{*{20}{c}} {{\rm{0}}{\rm{.554\, 6}}} & {{\rm{-0}}{\rm{.399\, 9}}} & {{\rm{-0}}{\rm{.293\, 1}}} \end{array}} \right]^{\rm T}}

    and

    {\dot e}\left( 0 \right) = \left[{\begin{array}{*{20}{c}} {{\rm{0}}{\rm{.071\, 5}}} & {{\rm{0}}{\rm{.104\, 7}}} & {{\rm{0}}{\rm{.060\, 9}}} \end{array}} \right].

    In addition, the target spacecraft is mobile and its angular rate \omega_r is

    {\omega _r} = 0.01\left[{\begin{array}{*{20}{c}} {\sin \left( {0.02t} \right)}\\ {-2\sin \left( {0.02t} \right)}\\ {\sin \left( {0.02t} \right)} \end{array}} \right]{\rm rad/s}.

    First of all, the controller {u} is designed by the approach in section 4 as

    \begin{align} & u={{u}_{c}}+{{u}_{f}}= \\ & \quad \ \ {{B}^{-1}}\left( \theta ,e,\dot{e} \right)\xi \left( \theta ,e,\dot{e},t \right)+ \\ & \quad \ \ {{K}_{0}}\left( \theta ,e,\dot{e} \right)e+{{K}_{1}}\left( \theta ,e,\dot{e} \right)\dot{e}+v. \\ \end{align}

    (62)

    Moreover, the controller should be optimized by the {{J} _{opt} } as

    \begin{array}{*{35}{l}} \min {{J}_{opt}}\left( F,Z \right) \\ \rm{s}.\rm{t}.\left\{ \begin{array}{*{35}{l}} \det \left( \begin{matrix} Z \\ ZF \\ \end{matrix} \right)\ne 0,~~Z\in {{\bf{R}}^{n\times 2n}} \\ {{\lambda }_{i}}\in {{C}^{-}},~~i=1,2,\cdots ,2n. \\ \end{array} \right. \\ \end{array}

    (63)

    Then, to simplify the optimization, the parameter {F} is considered as

    \begin{align} & F=\text{diag}\left( {{\lambda }_{1}},{{\lambda }_{2}},\cdots ,{{\lambda }_{6}} \right)= \\ & \quad \ \ \text{diag}\left( \text{-0}.\text{1},\text{-0}.\text{15},\text{-0}.\text{2},\text{-0}.\text{25},\text{-0}.\text{3},\text{-0}.\text{35} \right). \\ \end{align}

    The initial {Z} (0) is chosen by

    {Z}\left( 0 \right) = \left[{\begin{array}{*{20}{c}} 2 & 0 & 0 & 1 & 0 & 0\\ 0 & 2 & 0 & 0 & 1 & 0\\ 0 & 0 & 2 & 0 & 0 & 1 \end{array}} \right].

    Additionally, initial {{J} _{opt}} (0) is 17.551 4.

    Thus, the final optimal parameter {Z} is

    {Z} = {\left[{\begin{array}{*{20}{c}} {1.385\, 6} & {-0.001\, 1} & {0.000\, 3}\\ {0.001\, 3} & {1.474\, 2} & {-0.000\, 1}\\ {-0.000\, 6} & {0.000\, 0} & {1.335\, 9}\\ {1.199\, 1} & { - 0.000\, 9} & {0.001\, 2}\\ { - 0.000\, 4} & {1.163\, 1} & {0.000\, 0}\\ { - 0.000\, 2} & {0.000\, 1} & {1.285\, 9} \end{array}} \right]^{\rm{T}}}.

    Additionally, the final {{J} _{opt}} is 14.336 4.

    Note that

    {Z{F^2}{V^{ - 1}}} = {\left[{\begin{array}{*{20}{c}} {{\rm{-}}0.025\, 0} & {{\rm{-}}0.000\, 0} & {0.000\, 0}\\ {{\rm{-}}0.000\, 0} & {{\rm{ - }}0.045\, 0} & {{\rm{ - }}0.000\, 0}\\ {0.000\, 0} & {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.070\, 0}\\ {{\rm{ - }}0.350\, 0} & {{\rm{ - }}0.0001} & {0.000\, 1}\\ {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.450\, 0} & {{\rm{ - }}0.000\, 0}\\ {0.000\, 0} & {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.550\, 0} \end{array}} \right]^{\rm{T}}}.

    Because of the expression

    {V{V^{ - 1}}} = \left[{\begin{array}{*{20}{c}} {Z}\\ {ZF} \end{array}} \right]{V^{ - 1}} = \left[{\begin{array}{*{20}{c}} {Z{V^{-1}}}\\ {ZF{V^{-1}}} \end{array}} \right] = {I_6}

    (64)

    these equations can be derived as

    {{A_1}\left( {\theta, e, \dot e} \right)ZF{V^{ - 1}}} = \left[{\begin{array}{*{20}{c}} {{0_3}} & {{A_1}\left( {\theta, e, \dot e} \right)} \end{array}} \right]

    (65)

    {ZF{V^{ - 1}}} = \left[{\begin{array}{*{20}{c}} {{0_3}} & {{I_3}} \end{array}} \right]

    (66)

    \begin{align} & {{K}_{F}}=\left[ {{K}_{0}}~~~~{{K}_{1}} \right]=W{{V}^{-1}}= \\ & \quad \quad B{{\left( \theta ,e,\dot{e} \right)}^{-1}}\left( Z{{F}^{2}}+{{A}_{1}}\left( \theta ,e,\dot{e} \right)ZF \right){{V}^{-1}}= \\ & \quad \quad B{{\left( \theta ,e,\dot{e} \right)}^{-1}}\left( Z{{F}^{2}}{{V}^{-1}}+{{A}_{1}}\left( \theta ,e,\dot{e} \right)ZF{{V}^{-1}} \right)= \\ & \quad \quad B{{\left( \theta ,e,\dot{e} \right)}^{-1}}\left( Z{{F}^{2}}{{V}^{-1}}+\left[ {{0}_{3}}~~~~{{A}_{1}}\left( \theta ,e,\dot{e} \right) \right] \right). \\ \end{align}

    (67)

    Furthermore, with the controller u , the response of error attitude quaternion {{q}_{e}} = [{{q}_{e0}}~~ {{q}_{e1}}~~ {{q}_{e2}}~~ {{q}_{e3}}]^{\rm T} and the state variable {e} can be plotted in Matlab as Figs. 1 and 2.

    Figure  1.  Error attitude quaternion {q} _e
    Figure  2.  Response curve of variable {e} and \dot{e} of controlled system

    Based on Figs. 1 and 2, the variable {e} and \dot{e} are successfully made stable in time = 70 s, which reach the stable balance point [0, 0, 0]T. Thus, the controller {u} is successful for the attitude stabilization. According to (19), the error angular rate \omega_e also reached the zero point, which means the spacecraft successfully tracks the target with the angular rate {{ \omega_r }} = 0.01[\sin (0.02{t}), -2\sin (0.02{t}), \sin (0.02{t})] ^{\rm T} . According to (13) and (14), when the system is disturbed by external torque d , the error attitude quaternion q_e also reaches the stable equilibrium point [1, 0, 0, 0]T. Thus, the controlled spacecraft successfully tracks the mobile target. Moreover, the controller needs to be tested for availability and stability.

    Additionally, to justify the relative attitude quaternion between q_r for target spacecraft and q_d for controlled spacecraft, the tracking process is plotted by Matlab. Fig. 3 shows that when the simulation time gets near to 70 s, the controlled spacecraft attitude q_d successfully tracks on the target spacecraft attitude q_r .

    Figure  3.  Comparison between controlled spacecraft attitude and target spacecraft attitude

    According to Fig. 4, the controller is stable which is pointed out by the stable curve, and it can be utilized for the control model. Furthermore, there is a limitation of 2 N \cdot m for each axis based on that controlled spacecraft [3]. In Fig. 4, each axis control torque satisfies that requirement. Hence, the controller {u} is useful, practical, and has robustness eliminating the effects of the external disturbance torque.

    Figure  4.  Controller {u} with respect to time curve

    We verify whether the closed-loop system is a linear constant system as

    {\dot X }= {{A_c}X + {B_c}v}

    (68)

    where

    {A_c} = \left[{\begin{array}{*{20}{c}} 0 & 0 & 0 & 1 & 0 & 0\\ 0 & 0 & 0 & 0 & 1 & 0\\ 0 & 0 & 0 & 0 & 0 & 1\\ {{\rm{-}}0.025\, 0} & {{\rm{-}}0.000\, 0} & {0.000\, 0} & {{\rm{-}}0.350\, 0} & {{\rm{ - }}0.000\, 0} & {0.000\, 0}\\ {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.045\, 0} & {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.000\, 1} & {{\rm{ - }}0.450\, 0} & {{\rm{ - }}0.000\, 0}\\ {0.000\, 0} & {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.070\, 0} & {0.000\, 1} & {{\rm{ - }}0.000\, 0} & {{\rm{ - }}0.550\, 0} \end{array}} \right]

    {B_c} = \left[{\begin{array}{*{20}{c}} 0\\ {\dfrac{1}{2}{T}\left( {e} \right){J^{-1}}} \end{array}} \right].

    The state variable response curve of that linear constant system is plotted in Matlab by Fig. 5.

    Figure  5.  Response curve of variable {e} and \dot{e} of controlled system

    This linear constant system is stable as Fig. 5 describes. Although the state response of closed-loop linear system is similar to the stated response of variable {e} and { \dot{e}, } the state error between practical closed-loop and theoretical closed-loop linear constant systems need further test. Furthermore, the state error between practical closed-loop system and theoretical closed-loop linear constant system can be defined as

    state\_error\left( i \right) = {\left\| {{{\left[{\begin{array}{*{20}{c}} {e\left( i \right)} & {\dot e\left( i \right)} \end{array}} \right]}^{\rm T}} - {{\left[{X\left( i \right)} \right]}^{\rm T}}} \right\|_2}.

    (69)

    The response of the sate error is plotted in Fig. 6.

    Figure  6.  Curve of the state error between practical system and theoretical system

    According to Fig. 6, the state error between practical system and theoretical system is quite small which is of 10-14 level. Therefore, the controller {u} has successfully turned the original system into a linear constant stable system.

    A controller is designed by the direct parametric approach and applied on a practical spacecraft attitude system. In this paper, the nonlinear and non-approximated spacecraft attitude control model is turned into a stable closed-loop linear constant system by the controller {u} . Even though many other control strategies prefer to simplify the spacecraft attitude system with approximation, a fully-actuated second-order control model is built up without approximation. Furthermore, the values of the closed-loop eigenstructure are undetermined parameters. The user of the controller with all degrees of freedom {F} and {Z} can easily design the controller with desired closed-loop eigenstructure. Moreover, the degrees of freedom through the parameters {F} and {Z} are beneficial for the optimization of the controller to satisfy the requirement of performance of the controlled system, because the parameters can be uncertain before the simulation experiment. Additionally, the simulation by Matlab with a set of practical data for spacecraft attitude tracking on a mobile target verifies the availability and robust stability of the controller when there is the external disturbance torque applied on the control model. Finally, compared with the theoretical closed-loop linear constant system, the controller actually can convert the original system into a stable linear constant system, which means the controller is advantageous in easy placement of the desired stable closed-loop poles.

    The first author hereby extends her thanks to all the classmates in the Center for Control Theory and Guidance Technology in Harbin Institute of Technology, because of their help in the academic research and daily life. Additionally, the first author thanks professor Bin Zhou for his help of academic paper writing.

  • G. R. Duan. Quaternion-based satellite attitude controlA direct parametric approach. In Proceedings of the 14th International Conference on Control, Automation and Systems, IEEE, Seoul, Korea, pp. 935-941, 2014. http://www.nsfc.gov.cn/Portals/0/fj/fj20160106_01.xls
    G. R. Duan. Satellite attitude control-A direct parametric approach. In Proceedings of the 11th World Congress on Intelligent Control and Automation, IEEE, Shenyang, China, pp. 3989-3996, 2014. http://ieeexplore.ieee.org/xpl/abstractKeywords.jsp?reload=true&arnumber=7053383
    C. Y. Gao, Q. Zhao, G. R. Duan. Robust actuator fault diagnosis scheme for satellite attitude control systems. Journal of the Franklin Institute, vol. 350, no. 9, pp. 2560-2580, 2013. doi: 10.1016/j.jfranklin.2013.02.021
    A. G. Parlos, J. W. Sunkel. Adaptive attitude control and momentum management for large-angle spacecraft maneuvers. Journal of Guidance, Control and Dynamics, vol. 15, no. 4, pp. 1018-1028, 1992. doi: 10.2514/3.20937
    H. Bang, C. K. Ha, J. Hyoung Kim. Flexible spacecraft attitude maneuver by application of sliding mode control. Acta Astronautica, vol. 57, no. 11, pp. 841-850, 2005. doi: 10.1016/j.actaastro.2005.04.009
    S. N. Singh, R. Zhang. Adaptive output feedback control of spacecraft with flexible appendages by modeling error compensation. Acta Astronautica, vol. 54, no. 4, pp. 229-243, 2004. doi: 10.1016/S0094-5765(03)00030-4
    A. Tayebi. Unit quaternion-based output feedback for the attitude tracking problem. IEEE Transactions on Automatic Control, vol. 53, no. 6, pp. 1516-1520, 2008. doi: 10.1109/TAC.2008.927789
    D. Bustan, S. K. H. Sani, N. Pariz. Adaptive faulttolerant spacecraft attitude control design with transient response control. IEEE/ASME Transactions on Mechatronics, vol. 19, no. 4, pp. 1404-1411, 2014. doi: 10.1109/TMECH.2013.2288314
    R. Kristiansen, P. J. Nicklasson. Satellite attitude control by quaternion-based backstepping. In Proceedings of American Control Conference, IEEE, Portland, USA, vol. 2, pp. 907-912, 2005. https://www.infona.pl//resource/bwmeta1.element.ieee-art-000004554017
    D. Doody. Spacecraft attitude control. Deep Space Craft, D. Doody, Ed., Berlin Heidelberg, Germany:Springer, pp. 87-118, 2009.
    A. Grewal, V. J. Modi. Robust attitude and vibration control of the space station. Acta Astronautica, vol. 38, no. 3, pp. 139-160, 1996. doi: 10.1016/0094-5765(96)00073-2
    K. F. Lu, Y. Q. Xia. Adaptive attitude tracking control for rigid spacecraft with finite-time convergence. Automatica, vol. 49, no. 12, pp. 3591-3599, 2013. doi: 10.1016/j.automatica.2013.09.001
    A. L. Herman, B. A. Conway. Optimal spacecraft attitude control using collocation and nonlinear programming. Journal of Guidance, Control and Dynamics, vol. 15, no. 5, pp. 1287-1289, 1992. doi: 10.2514/3.20983
    B. Wie, K. W. Byun, V. W. Warren, D. Geller, D. Long, J. Sunkel. New approach to attitude/momentum control for the Space Station. Journal of Guidance, Control and Dynamics, vol. 12, no. 5, pp. 714-722, 1989. doi: 10.2514/3.20466
    J. D. Schierman, D. G. Ward, J. R. Hull, N. Gandhi, M. Oppenheimer, D. B. Doman. Integrated adaptive guidance and control for re-entry vehicles with flight test results. Journal of Guidance, Control and Dynamics, vol. 27, no. 6, pp. 975-988, 2004. doi: 10.2514/1.10344
    H. J. Shen, P. Tsiotras. Optimal two-impulse rendezvous using multiple-revolution lambert solutions. Journal of Guidance, Control and Dynamics, vol. 26, no. 1, pp. 50-61, 2003. doi: 10.2514/2.5014
    F. Zhang, G. R. Duan. Optimal integrated relative position and attitude control of spacecraft in proximity operation missions. International Journal of Automation and Computing, vol. 9, no. 4, pp. 342-351, 2012. doi: 10.1007/s11633-012-0654-0
    H. Du, G. L. Fan, J. Q. Yi. Nonlinear longitudinal attitude control of an unmanned seaplane with wave filtering. International Journal of Automation and Computing, vol. 13, no. 6, pp. 634-642, 2016. doi: 10.1007/s11633-016-0962-x
    W. K. Liu. Research on Attitude Maneuver and Tracking for Flexible Spacecraft, Master dissertation, Harbin Institute of Technology, China, 2014. (in Chinese)
    L. Zhang, G. R. Duan, Y. A. Zhang. Nonlinear dynamic model and control of three-axis stabilized liquid-filled spacecraft attitude system. Journal of Harbin Institute of Technology (New Series), vol. 19, no. 5, pp. 107-112, 2012. http://www.cqvip.com/QK/86045X/201205/43848684.html
    G. R. Duan. Direct parametric control of fully-actuated second-order nonlinear systems-The normal case. In Proceedings of the 33rd Chinese Control Conference, IEEE, Nanjing, China, pp. 2406-2413, 2014. http://www.nsfc.gov.cn/Portals/0/fj/fj20160106_01.xls
    G. R. Duan, B. Zhou. Solution to the second-order Sylvester matrix equation MVF2+ DVF + KV=BW. IEEE Transactions on Automatic Control, vol. 51, no. 5, pp. 805-809, 2006. doi: 10.1109/TAC.2006.874989
    G. R. Duan. Generalized Sylvester Equations:Unified Parametric Solutions, New York, USA:CRC Press, pp. 15-393, 2015.
    G. R. Duan. Parametric eigenstructure assignment in second-order descriptor linear systems. IEEE Transactions on Automatic Control, vol. 49, no. 10, pp. 1789-1794, 2004. doi: 10.1109/TAC.2004.835580
  • Related Articles

    [1]Zhen-Guo Liu, Jin-Ming Huang. A New Adaptive Tracking Control Approach for Uncertain Flexible Joint Robot System[J]. Machine Intelligence Research, 2015, 12(5): 559-566. DOI: 10.1007/s11633-015-0898-6
    [2]Wen-Rui Hu, Yuan Xie, Lin Li, Wen-Sheng Zhang. A TV-l1 Based Nonrigid Image Registration by Coupling Parametric and Non-parametric Transformation[J]. Machine Intelligence Research, 2015, 12(5): 467-481. DOI: 10.1007/s11633-014-0874-6
    [3]Qian Wang, Guang-Ren Duan. Robust Global Stabilization of Spacecraft Rendezvous System via Gain Scheduling[J]. Machine Intelligence Research, 2014, 11(4): 426-433. DOI: 10.1007/s11633-014-0809-2
    [4]Pan Wang, Wei-Wei Sun. Adaptive H Control for Nonlinear Hamiltonian Systems with Time Delay and Parametric Uncertainties[J]. Machine Intelligence Research, 2014, 11(4): 368-376. DOI: 10.1007/s11633-014-0802-9
    [5]Zhi-Qiang Pu, Ru-Yi Yuan, Xiang-Min Tan, Jian-Qiang Yi. An Integrated Approach to Hypersonic Entry Attitude Control[J]. Machine Intelligence Research, 2014, 11(1): 39-50. DOI: 10.1007/s11633-014-0764-y
    [6]Feng Zhang, Guang-Ren Duan. Integrated Relative Position and Attitude Control of Spacecraft in Proximity Operation Missions[J]. Machine Intelligence Research, 2012, 9(4): 342-351. DOI: 10.1007/s11633-012-0654-0
    [7]Li-Jie Zhao, Chang-Ping Tang, Peng Gong. Correlation of Direct Piezoelectric Effect on EAPap under Ambient Factors[J]. Machine Intelligence Research, 2010, 7(3): 324-329. DOI: 10.1007/s11633-010-0510-z
    [8]Shu-Xi Liu, Ming-Yu Wang, Yu-Guang Chen, Shan Li. A Novel Fuzzy Direct Torque Control System for Three-level Inverter-fed Induction Machine[J]. Machine Intelligence Research, 2010, 7(1): 78-85. DOI: 10.1007/s11633-010-0078-7
    [9]Bin Zhou, Guang-Ren Duan, Yun-Li Wu. Parametric Approach for the Normal Luenberger Function Observer Design in Second-order Descriptor Linear Systems[J]. Machine Intelligence Research, 2008, 5(2): 125-131. DOI: 10.1007/s11633-008-0125-9
    [10]Akira Inoue, Ming-Cong Deng. Framework of Combined Adaptive and Non-adaptive Attitude Control System for a Helicopter Experimental System[J]. Machine Intelligence Research, 2006, 3(3): 229-234. DOI: 10.1007/s11633-006-0229-z
  • Cited by

    Periodical cited type(2)

    1. Xiaoyi Wanga, Lingling Shi, Jayantha Katupitiya. Coordinated Control of a Dual-arm Space Robot to Approach and Synchronise with the Motion of a Spinning Target in 3D Space. Acta Astronautica, 2020. DOI:10.1016/j.actaastro.2020.02.028
    2. Zhanyu Gao, Yingying Gu, Yaoyu Lv, et al. A Monocular Vision Measurement System of Three-Degree-of-Freedom Air-Bearing Test-Bed Based on FCCSP. Photonic Sensors, 2018, 8(2): 119. DOI:10.1007/s13320-018-0420-1

    Other cited types(0)

Catalog

    Figures(6)

    Scan QR codes using WeChat, hare with friends and Moments

    Article Metrics

    Article views (820) PDF downloads (23) Cited by(2)

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return