An Optimal Fuzzy PID Control Approach For Docking

Engineering Science and Technology, an International Journal xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Engineering Science and Technology,

an International Journal
journal homepage: www.elsevier.com/locate/jestch
Full Length Article
An optimal fuzzy PID control approach for docking maneuver of two

spacecraft: Orientational motion
A. Kosari ⇑, H. Jahanshahi, S.A. Razavi
Faculty of New Sciences and Technologies (FNST), University of Tehran, Tehran, Iran
a r t i c l e i n f o a b s t r a c t
Article history: This paper describes a scheme for a Fuzzy-Proportional Integral Derivative (FPID) controller based on
Received 30 April 2016 genetic algorithm (GA), in a docking maneuver of two spacecraft. The docking maneuver consists of
Revised 29 June 2016 two parts: translation and orientation. Euler’s gyroscopic equation is applied to obtain governing equa-
Accepted 30 July 2016
tions of orientational phase. Here, a designed fuzzy-PID controller for stabilization purpose of orienta-
Available online xxxx
tional phase of a docking maneuver is presented based on the Single Input Fuzzy Inference Motor
(SIFIMs) dynamically connected Preferrer Fuzzy Inference Motor (PFIM). This fuzzy-PID controller takes
Keywords:
the error signal of Euler’s angles and the error of angular velocities of the chaser as its input items, and the
Fuzzy-PID
Genetic algorithm
driving force as its output. The parameters of the controller are ascertained by using a genetic algorithm.
Docking maneuver Conflicting objective functions (which their 3D pareto frontiers are obtained by Multi-objective Genetic
Pareto frontier Algorithm (MOGA)) are distance errors from the set point, angle errors from the set point, and control
Orientation motion efforts. Optimization constraint is maximal of the momentum produced by momentum wheels. The
Euler gyroscopic equation result of optimum point demonstrates that the designed controller makes an efficient performance in
Momentum wheel the orientational phase of the chaser spacecraft. Compared to similar works, some of system parameters
like settling time are improved and overshoot (as a critical parameter in docking maneuver) is decreased.
Ó 2016 Karabuk University. Publishing services by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction strategies have been applied to Autonomous Underwater Vehicle

(AUV) docking task. [6–8]. Villagra and Herrero-Pérez [9] presented
The rendezvous and docking (RVD) process consists of a series a comparison in detail between different control approaches for
of orbital maneuvers and controlled trajectories, which eventually Automated Guided Vehicle (AGV) robust path tracking.
dock the chaser with the target. The term docking is used for the Poor attitude control in docking maneuver can definitely cause
case where the Guidance, Navigation and Control (GNC) system mission failure, so precise controller is required. In spite of novelty
of the chaser controls vehicle state parameters to ensure entering of docking and rendezvous problem in space technology, some new
its capturing interfaces into those of the target vehicle, and the control approaches have been implemented. Ho and McClamroch
capture location is the location for structural connection [1]. For [10] proposed an automatic spacecraft docking control system
an actively-maneuvering spacecraft, a docking mechanism is used using a computer vision system as relative position and orientation
for joining to other spacecraft to constitute an integral unit. Recov- sensors. Park et al. [11] suggested a model predictive control
ering a tumbling satellite, Capturing the target satellite and con- approach for spacecraft docking with tumbling platform. Introduc-
trolled re-entry into atmosphere, performing repair, in-orbit ing a novel control algorithm for close-proximity multiple-
servicing of low Earth orbit satellites, multiple docking of space- spacecraft autonomous maneuvers in docking is done by [2]. Xia
craft capsules for extending the mission objectives, makes the and Huo [12] investigated a robust adaptive backstepping neural
docking maneuver a necessity in space missions [1–3]. networks control strategy for rendezvous and docking of spacecraft
Docking task has a wider domain of applications than that per- considering coupled position and attitude dynamics. Filipe and
taining to space. Docking plays an important role for any mobile Tsiotras [13] proposed a global asymptotically stable nonlinear
robot, looks for interaction with objects [4,5]. Various control adaptive position and attitude controller for satellite proximity
operations using dual quaternion.
Fuzzy systems are knowledge-based or rule-based systems
⇑ Corresponding author. formed via human knowledge and heuristics. They have been
E-mail address: Kosari_a@ut.ac.ir (A. Kosari). applied to a wide range of fields such as control, communication,
Peer review under responsibility of Karabuk University.
http://dx.doi.org/10.1016/j.jestch.2016.07.018
2215-0986/Ó 2016 Karabuk University. Publishing services by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article in press as: A. Kosari et al., An optimal fuzzy PID control approach for docking maneuver of two spacecraft: Orientational motion,
Eng. Sci. Tech., Int. J. (2016), http://dx.doi.org/10.1016/j.jestch.2016.07.018
2 A. Kosari et al. / Engineering Science and Technology, an International Journal xxx (2016) xxx–xxx
Fig. 1. Schematic of designed controller structure.
medicine, management, business, psychology, etc. The most signif- controllers for various engineering problems has been a major
icant applications and studies about fuzzy systems have concen- research activity in recent years. Different kind of studies have
trated on the control area [14–24]. The development of fuzzy-PID been done in applying fuzzy logic to docking maneuver problem.
A guidance control method based on fuzzy logic system is pro-
posed for the mission of autonomous rendezvous and docking with
non-cooperative target spacecraft [25]. Leitner et al. [26] designed
a neurocontroller to reproduce the optimal control for an auto-
matic rendezvous and docking task.
The heuristic parameters of fuzzy-PID controllers have to be
determined via an appropriate approach. A very effective way to
choose these factors is use of evolutionary algorithms, such as
genetic algorithm (GA). A constrained optimization of a simple
fuzzy-PID system is designed for the online improvement of PID
control performance during productive control runs [27]. Duan
et al. [28] proposed an inherent saturation of the fuzzy-PID con-
troller revealed due to the finite fuzzy rules. Oh et al. [29] devel-
oped a design methodology for a fuzzy PD cascade controller for
a ball-beam system using particle swarm optimization (PSO). An
on-line tuning method is proposed for fuzzy PID controllers via
rule weighing [30]. Boubertakh et al. [31] proposed a new auto-
tuning fuzzy PD and PI controllers using reinforcement-learning
(QL) algorithm for SISO (single-input single-output) and TITO
(two-input two-output) systems. Nie and Tan [32] presented an
improved version of the stable fuzzy adaptive control structure,
Table 1
Angular mass of chaser spacecraft.
Variable Value Description
JCxx 2000 Chaser angular mass around x ½kg=m2

JCyy 5000 Chaser angular mass around y ½kg=m2
JCzz 2000 Chaser angular mass around z ½kg=m2
Fig. 2. RVD mission elements breakdown.
Fig. 3. Chaser and target spacecraft: before and after orientation phase.
A. Kosari et al. / Engineering Science and Technology, an International Journal xxx (2016) xxx–xxx 3
Fig. 4. The block diagram of the optimal fuzzy-PID control.
which comprises an approximation of the ideal controller and a

supervisory controller. Mahmoodabadi and Jahanshahi [33] stud- Table 3
ied multi-objective optimization algorithms for the optimum The rules of SIFIMs for the spacecraft system.
design of the fuzzy PID controller for two nonlinear benchmarks.
If Then
X i ði ¼ 1; 2; 3; 4; 5; 6; 7; 8; 9Þ
VBi f r1 ¼ 1
Table 2 POi f r2 ¼ 0
The membership functions of SIFIM for the spacecraft system. ZBi f r3 ¼ 1
If Then
X i 6 1 VBi ¼ 1
Table 4
POi ¼ 0
The membership functions of PFIM for the spacecraft system.
ZBi ¼ 0
1 6 X i 6 0 VBi ¼ X i If Then
POi ¼ X i þ 1
ZBi ¼ 0 jX 2 j 6 0:5 HS ¼ 2jX 2 j þ 1
0 6 Xi 6 1 VBi ¼ 0 HM ¼ 2jX 2 j
POi ¼ X i þ 1 HB ¼ 0
ZBi ¼ X i 0:5 6 jX 2 j HS ¼ 0
1 6 Xi VBi ¼ 0 HM ¼ 2jX 2 j þ 2
POi ¼ 0 HB ¼ 2jX 2 j 1
ZBi ¼ 1 jX 5 j 6 0:5 HS ¼ 2jX 5 j þ 1
HM ¼ 2jX 5 j
HB ¼ 0
0:5 6 jX 5 j HS ¼ 0
HM ¼ 2jX 5 j þ 2
HB ¼ 2jX 5 j 1
jX 8 j 6 0:5 HS ¼ 2jX 8 j þ 1
HM ¼ 2jX 8 j
HB ¼ 0
0:5 6 jX 8 j HS ¼ 0
HM ¼ 2jX 8 j þ 2
HB ¼ 2jX 8 j 1
Fig. 5. Membership functions for each SIFIM.
formed to state space form. Here, SIFIMs dynamically connected

PFIM are utilized as two fuzzy inference motors. The parameters
of the controller are ascertained by using a genetic algorithm. 3D
Pareto-frontiers of conflicting objective functions (which obtained
by MOGA) are distance errors from the set point, angle errors from
the set point, and control efforts. Optimization constraint is maxi-
Fig. 6. Membership functions for each PFIM. mal of the momentum produced by momentum wheels. The result
of optimum point demonstrates that the designed controller makes
an efficient performance in the costs of orientational phase of the
Table 5 chaser spacecraft and could satisfy the mission time. Finally, com-
The rules of PFIM for the spacecraft system.
pared to similar works, priority of our work in some of system per-
If Then formance parameters (e.g. settling time and overshoot) are
jX 2 j HS W1 ¼0 discussed.
jX 2 j HM W2 ¼ 0:5
jX 2 j HB W3 ¼1
2. Optimal fuzzy-PID controller design
jX 5 j HS W4 ¼0
jX 5 j HM W5 ¼ 0:5
jX 5 j HB W6 ¼1 In this section, the optimal fuzzy-PID controller is designed for
jX 8 j HS W7 ¼0 the chaser spacecraft. For orientational motion of chaser and target
jX 8 j HM W8 ¼ 0:5 spacecraft, fuzzy-PID controller would be implemented. Fig. 1
jX 8 j HB W9 ¼1
demonstrates a schematic view of designed controller structure
in this work. As it shown, outputs are compared with their desired
In their work an integral term was augmented to the state vari- input values and error function which is the input of fuzzy and PID
ables in order to eliminate the steady state errors and decrease controllers is made. Base and regulation variables obtained by
the rising time. Sahib [34] proposed an optimized proportional, genetic algorithm in h—u—w directions and the error function,
integral, derivative and second order derivative order (PIDD2) are considered as fuzzy controller inputs. Fuzzy controller calcu-
using particle swarm optimization for automatic voltage regulator lates integral-proportional-derivative coefficients of PID controller
(AVR). Ortega and Giron-Sierra [35] studied the use of Genetic and with multiplying these values by integral-proportional-deriva
Algorithms (GAs) to perform the optimization of the fuzzy con- tive of input error by PID controller, final control force is made.
troller by finding the best fuzzy sets of the membership functions,
to optimize docking time and fuel consumption. 2.1. Orientation
In this paper, the novel optimal Fuzzy-PID control strategy for
orientational part of a docking maneuver of two spacecraft is pro- At first step, the desired mission and required systems is men-
posed. To derive governing equations for orientational phase of tioned, then dynamics of space vehicles for successfully accom-
docking, Euler’s gyroscopic equation is used which then trans- plishing of noticed mission is studied. RVD mission can break
Fig. 7. 3-D pareto front by objectives 1, 2, and 3 of optimal fuzzy-PID for orientational phase of chaser spacecraft.
down into several components and sub-components. Rendezvous range rendezvous operation. Suitable navigation during ren-
operation providing conditions for docking task, splits up into the dezvous (i.e. far range or close range rendezvous) is achieved
two parts: based on direction and range relative measurements or directly
by relative position.
Far range Rendezvous operation whose main purpose is reduc- Close range rendezvous is usually divided into two subphases: a
tion of trajectory dispersions and achievement of proper posi- preliminary phase providing conditions for closing and a final
tion, velocity and angular rate conditions for initiation of close one which leads to the mating. In the closing phase, reduction
of range to the target and placing the chaser in suitable place for At the end of final approach phase, docking operation begins.
accomplishing the final phase of mission, is considered. The Mating phase starts when the chaser GNC system, has delivered
main purpose of final approach is acquiring proper berthing or the capture interfaces of the chaser into the right range to the tar-
docking conditions. get. Capturing target by chaser is the final phase of RVD operation.
Fig. 2 shows the RVD operation from launch to capturing phase.
By the end of rendezvous operation, two spacecraft stand in a The initial and desired values are x ¼ ½x1 ; x2 ; x3 ; x4 ; x5 ; x6 ; x7 ;
position like Fig. 3a. Docking operation consists of two phases x8 ; x9 ¼ ½0; 0; 0; 0; 0; 3:0023 ; 0; 200 ; 0 and x ¼ ½x1 ; x2 ; x3 ; x4 ; x5 ;
including orientation and translation. Firstly, chaser spacecraft x6 ; x7 ; x8 ; x9 ¼ ½0; 0; 0; 0; 0; 0; 0; 0; 0, respectively. The chaser angu-
should perform a rotational motion to place its port along with lar mass around x; y and z axes are listed in Table 1.
the port of target spacecraft. At the end of orientation phase, two As shown in Fig. 4 the block diagram of the optimal Fuzzy-PID
spacecraft are sit in the position as Fig. 3b. After this operation, control (for stabilization control of orientation of the chaser space-
R
only a translation phase is required for the final docking. craft system) represents each of state variables h; h; xSx ;
R R
Now, the orientation equation of chaser is considered. Repre- /; /; xy ; w; w; xz relevant to the orientation of the chaser space-
S S
senting the inertia tensor with J, Euler’s gyroscopic equation craft system is fed back and compared with its desired value.
becames as Eq. (1): Two fuzzy inference motors are utilized in the suggested
J_ x þ þJ x
_ þ x ðJ xÞ ¼ m ð1Þ Fuzzy-PID controller. The first one called Single Input Fuzzy
Inference Motor (SIFIM). This motor has one input with a
There are two important assumptions taken: constant mass dis- separate SIFIM defined for each state variable (9 SIFIMs) and
tribution over time, i.e., J_ ¼ 0 and the coincidence of system with each input item X i ; i ¼ 1; 2; 3; 4; 5; 6; 7; 8; 9 is fed into the corre-
the principal axes of the body (J ik ¼ 0 for all i–k). The state vector sponding SIFIM-i. The second fuzzy inference motor is Preferrer
is the system observable state vector x ¼ ½x1 ; x2 ; x3 ; x4 ; x5 ; x6 ; x7 ; x8 ; Fuzzy Inference Motor (PFIM) that is used for the control priority
R R R
x9 ¼ ½ h; h; xSx ; u; u; xSy ; w; w; xSz . The state variables x2 , x5 order of each state variable. For X i ; i ¼ 1; 2; 3 PFIM blocks take
and x8 denote Euler xyz-angles (h, u and w respectively). The state the absolute values of the input item X 2 as their antecedent
variables x1 , x4 and x7 are integrals of the state variables x2 , x5 and variables. For X i ; i ¼ 4; 5; 6 PFIM blocks do similar toward input
x8 , respectively. The additional state variables x3 , x6 and x9 are the item X 5 , and PFIM blocks for X i ; i ¼ 7; 8; 9 do the same for input
respective angular velocities transforming the system into the nine item X 8 .
first order differential Eq. (2):
x_ 1 ¼ x2
x_ 2 ¼ x3 Table 6
x_ 3 ¼ J1S x6 x9 J Syy J Szz þ mx Design variables for the optimum design point A.
xx
x_ 4 ¼ x5 Design Value Design Value Design Value

x_ 5 ¼ x6 ð2Þ variable variable variable
x_ 6 ¼ J1S x3 x9 J Szz J Sxx þ my K bih 0:00010 K biu 0:00495 K biw 0:00002

yy
K bh 0:97563 K bu 0:08541 K bw 0:000171
x_ 7 ¼ x8
K bdh 0:20740 K bdu 0:99996 K bdw 0:19493
x_ 8 ¼ x9 K rih 0:14633 K riu 0:09743 K riw 0:00233
x_ 9 ¼ J1S x3 x6 J Sxx J Syy þ mz K rh 0:03892 K ru 1:21991 K rw 0:12165
zz
K rdh 0:00048 K rdu 0:00126 K rdw 0:07167

where m ¼ ½ mx my mz T describes the momentum vector.
The membership functions of SIFIMs for spacecraft system are For the spacecraft system DW 1 ; DW 2 ; DW 3 ; DW 4 ; DW 5 ; DW 6 ;
shown in Table 2 and Fig. 5. The rules of the SIFIMs are as Table 3. DW 7 ; DW 8 and DW 9 (The outputs of SIFIM-i), are obtained from
For the spacecraft system, fi, the output of SIFIM-i, is calculated Eq. (4).
from Eq. (3).
DW 1 ¼ DW 2 ¼ DW 3 ¼ DW 4 ¼ DW 5 ¼ DW 6 ¼ DW 7 ¼ DW 8 ¼ DW 9
VBi f r1 þ POi f r1 þ ZBi f r3 W 1 HS þ W 2 HM þ W 3 HB þ W 4 HS þ W 5 HM
fi ¼ ð3Þ ¼
VBi þ POi þ ZBi HS þ HM þ HB
þW 6 HB þ W 7 HS þ W 8 HM þ W 9 HB
The membership functions of PFIMs for the spacecraft system ð4Þ
HS þ HM þ HB
are shown in Table 4 and Fig. 6. The rules of the PFIMs are tabu-
lated in Table 5. Given f i and DW i , it is likely to introduce the fuzzy-PID con-
troller as Eq. (5):
Z ^ Z ^
Table 7 ^ ih
f ¼K ^hdt þ K ^ dh dh þ K
^ ph ^h þ K ^ i/ /d ^ p/ /
^ t þK ^ d/ d/
^ þK
Objective functions for the optimum design points. dt dt
Z ^
Point Objective Value Point Objective Value ^ iw wd ^ t þK ^ pw w
^ þK ^ dw dw
þK ð5Þ
function function dt
A O:F:1 272:151 B O:F:1 31:799
where f is the control action;
O:F:2 393:775 O:F:2 127047:491 R ^ R R
O:F:3 5746:200 O:F:3 59741:050 hdt ; ^
^ h; ddht ; u
^ dt ; u ^ and dw^ are the fuzzy forms of
^ t; w
^ ; ddu^ ; wd dt
O:F:4 14:083 O:F:4 39:553 R R R t
O:F:5 0:298 O:F:5 198:717

hdt ; h; dh
dt
; udt ; u; ddut ; wdt ; w and dw dt
, respectively which are
R
O:F:6 26:146 O:F:6 1:482 obtained by SIFIM. We can also write the following: hd ^ t¼f ;
1
C O:F:1 268:065 D O:F:1 298:327 R R
^ ^
h ¼ f 2 ; ddht ¼ f 3 ; u^ dt ¼ f 4 ; u u^
^ ¼ f 5; d ¼ f 6;
d ^ t ¼f ;
wd w^¼
O:F:2 380:941 O:F:2 417:887 t 7
O:F:3 5955:862 O:F:3 5294:687 ^
dw
f 8; dt
¼ f 9 . Moreover, K bih , K bh , K bdh , K biu , K bu , K bdu , K biw , K bw and K bdw
O:F:4 7:327 O:F:4 32:386
O:F:5 16:368 O:F:5 6:164 denote the fuzzy variables found by the Eqs. (6)–(14):
O:F:6 2:5860 O:F:6 102:124
^ ih ¼ K b þ K r DW 1
K ð6Þ
E O:F:1 8355:615 F O:F:1 306:436 ih ih
O:F:2 42532:986 O:F:2 389:495
O:F:3 37501:744 O:F:3 5731:320 ^ h ¼ K b þ K r DW 2
K ð7Þ
h h
O:F:4 0:075 O:F:4 16:034
O:F:5 1600:902 O:F:5 0:094
O:F:6 342:307 O:F:6 66:714 ^ dh ¼ K b þ K r DW 3
K ð8Þ
dh dh
G O:F:1 85:293
O:F:2 122837:628
^ i/ ¼ K b þ K r DW 4
K ð9Þ
O:F:3 59876:552 i/ i/
O:F:4 23:046
O:F:5 649:427 ^ u ¼ K b þ K r DW 5
K ð10Þ
O:F:6 1:248 u u
Fig. 17. The position of chaser spacecraft in h-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
^ d/ ¼ K b þ K r DW 6
K ð11Þ
d/ d/ Derivatives of function are not concerned in this method, how-
ever the most of deterministic methods involve different order
^ iw ¼ K b þ K r DW 7
K ð12Þ
iw iw derivatives. Therefore, in this paper, genetic algorithm method is
chosen for finding optimal control variables.
^ w ¼ K b þ K r DW 8
K ð13Þ
w w
3. Results and discussion

^ dw ¼ K b þ K r DW 9
K ð14Þ
dw dw
Here, the multi-objective optimization of the proposed Fuzzy-
in which K bih , K bh , K bdh , K biu , K bu , K bdu , K biw , K bw and K bdw denote the base PID controller would be done with respect to nine design variables
variables and K rih , K rh , K rdh , K ri/ , K r/ , K rd/ , K riw , K rw and K rdw denote the reg- and six objective functions. The base values K bih ; K bh ; K bdh ; K bi/ ;
ulation variables. Usually, the base and regulation variables can be K b/ ; K bd/ ; K biw ; K bw and K bdw and regulation values K rih ; K rh ; K rdh ; K ri/ ; K r/ ;
found by trial and error.
K rd/ ; K riw ; K rw and K rdw are the design variables. The angular error of
Compared with other standard optimization methods, Genetic
algorithm has important superiorities: the spacecraft in the h—u—w of ‘‘Euler-axis/angle” description and
the control effort in corresponding directions are the objective func-
Parallel computing is one of the main features of genetic algo- tions. In summary, the objective functions are as Eqs. (15)–(20):
Z
rithm, i.e. in this method in a certain time a population is moved
O:F:1 ¼ jxjdt ð15Þ
toward optimal point rather than a single variable. As a result,
convergence speed of this method increases significantly.
Z
In this method, one can optimize non-smooth behavior prob-
O:F:2 ¼ jyjdt ð16Þ
lems (e.g. functions with cyclic periods which have lots of rela-
tive minima or functions with highly nonlinearity) with Z
acceptable scale.
O:F:3 ¼ jzjdt ð17Þ
It’s an efficient tool for the problems dealing with discrete
variables. Z
O:F:4 ¼ ju1 jdt ð18Þ
Table 8
Output of proposed controller of the point A in directions of Euler angles.
Z
Direction Settling Maximum Maximum Maximum O:F:5 ¼ ju2 jdt ð19Þ
time (s) overshoot angular velocity driving force
(deg) (deg/s) (N m) Z
h 150 5.9 3.6 0.52 O:F:6 ¼ ju3 jdt ð20Þ
/ 120 0.35 1.7 0.005
w 212 113 23.7 0.65
Fig. 18. The angular velocity of chaser spacecraft in h-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 19. The driving momentum of chaser spacecraft in h-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 20. The position of chaser spacecraft in /-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 21. The angular velocity of chaser spacecraft in /-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 22. The driving momentum of chaser spacecraft in /-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 23. The position of chaser spacecraft in w-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
Fig. 24. The angular velocity of chaser spacecraft in w-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
In this problem, the algorithm configuration of the genetic algo- C, D, E, F and G have infimum values in directions of objective func-
rithm is as follows. The crossover fraction = 0.8, population size = tions 2, 3, 4, 5 and 6 axes. These are the best points for objectives 2,
500, selection function = tournament, mutation function = 3, 4, 5 and 6 among all the output points of optimization. By adding
constraint dependent, crossover function = intermediate, crossover objectives 1 to 6 with equal weighting factors, point A is the best
ratio = 1, migration direction = forward, migration fraction = 0.2, point where its corresponding design variables, controls system
migration interval = 20, distance measure function = distance with minimum possibly overshoot and settling time.
crowding, Pareto front population function = 0.35, and stopping Angular position of chaser spacecraft in direction of h is shown
criteria is defined as function tolerance = 104. in Fig. 17 for all the points of A, B, C, D, E, F and G. Design variables
To compare and make a precise view of objective functions, 3D related to point E as an optimum point to the objective 5, provide a
pareto frontiers by different conflicting objective functions for the controller which is not able to control system in directions of h, /
points A, B, C, D, E, F and G are shown in Figs. 7–16. and w, However, points A, C, D and F as design points causes fully
Design variables for the optimum design point A and objective control of system in all directions. By evaluating control effort of
functions for the optimum design points are given in Tables 6 and these four design points, points A and B have the minimum and
7, respectively. maximum control effort, respectively. The value of settling time,
Complete stabilization occurs where all the state variables con- maximum overshoot, maximum angular velocity and maximum
verge to zero. Settling time for optimum point A in the orientation required control effort for point A in directions of h, / and w is
task of docking manoeuvre is 360 s, and the maximum driving shown in Table 8 (see Figs. 18–25).
momentum is around 0.752 N m. Michael et al. [36] have reported the same results. Their work is
Among all the optimum points as outputs of GA, point B has the similar to this paper except the point that the chaser is in a non-
infimum value in direction of the axis namely objective function 1. rotated initial state and its angular velocity is zero upon start of
This is the best point for objective 1 in all the figures. The points B, the manoeuvre. Unlike the chaser, the target also starts in a
Fig. 25. The driving momentum of chaser spacecraft in w-direction for optimum design points A, B, C, D and E illustrated in the pareto front.
φT
100 θT
Angle (deg)
ΨT
0
φS
-100 θS
ΨS
50 100 150 200 250 300 350 400
Time (s)
Fig. 26. Orientation of target and chaser (dashed).
wxT
Angular Velocity (rad/s)

0.04
wyT
0.02 wzT
wSx
0 wSy
wSz
-0.02
50 100 150 200 250 300 350 400
Time (s)
Fig. 27. Angular velocity of target and chaser (dashed).
0.04
Momentum (Nm)
mx
0.02
my
mz
0
-0.02
50 100 150 200 250 300 350 400
Time (s)
Fig. 28. Momentum control of target and chaser system.
non-rotating initial state and rotates with a constant angular veloc- novel optimal fuzzy-PID controller, chaser spacecraft would fulfill
ity of 3 =s around its y axis which leads to a stable motion with its task in less possible time and minimum propellant consump-
respect to the time. The assumptions of initial values are as tion. Generally, in this work, performing docking maneuver with
follows: the least control effort, accurate pointing and settling time
2 2 3 2
3 3 2 3 2 3 2 3 decrease was considered in the purpose of controller design pro-
x0 0 v x0 0 0 0
cess. The reported result indicated that expected controller can
6 7 6 7 6 7 6 7 6 7 6 7
4 y0 5 ¼ 4 3 5; 4 v y0 5 ¼ 4 0 5; xS ¼ 4 0 5; xT ¼ 4 0:052395ðradÞ 5 effectively response well to the given constraints and conditions.
z0 0 v z0 0 0 0 In comparison with similar works, some of system parameters like
settling time are improved, overshoot (as a critical parameter in
To make a real comparison, output results of Michael‘s work for soft-docking maneuver) is decreased and the less amount of
Orientation, angular velocity and momentum control of target and energy is required.
chaser is demonstrated in Figs. 26–28, respectively.
Although there are some differences between our work and
Michael et al. [36] in initial conditions assumptions for target References
and chaser, the objective of research is the same. By comparison
of these two works, superiority of our work from viewpoints of dis- [1] W. Fehse, Automated Rendezvous and Docking of Spacecraft, vol. 16,
tance error, angle error and control effort decrease is distinct. Cambridge University Press, 2003.
[2] R. Bevilacqua, T. Lehmann, M. Romano, Development and experimentation of
In orientation task, as shown in Fig. 26, Michael’s work focus on LQR/APF guidance and control for autonomous proximity maneuvers of
a tracking problem. It reached to the final state almost in 270 s, multiple spacecraft, Acta Astronaut. 68 (2011) 1260–1275.
360 s and 270 s, in h, / and w directions, respectively, while we [3] N. Philip, M. Ananthasayanam, Relative position and attitude estimation and
control schemes for the final phase of an autonomous docking mission of
can achieve it almost 150 s in h direction, 120 s in / direction spacecraft, Acta Astronaut. 52 (2003) 511–522.
and 212 s in w direction, which is far better. As shown in Fig. 28, [4] C. McCarthy, N. Barnes, R. Mahony, A robust docking strategy for a mobile
obviously, there is no off time for momentum wheels in Michael’s robot using flow field divergence, IEEE Trans. Robotics 24 (2008) 832–842.
[5] C. McCarthy, N. Barnes, A unified strategy for landing and docking using
work and they need more energy to run compared to our work. spherical flow divergence, IEEE Trans. Pattern Anal. Mach. Intell. 34 (2012)
1024–1031.
[6] K. Teo, B. Goh, O.K. Chai, Fuzzy docking guidance using augmented navigation
4. Conclusion system on an AUV, IEEE J. Oceanic Eng. 40 (2015) 349–361.
[7] L. Brignone, M. Perrier, C. Viala, A fully autonomous docking strategy for
In this study, designing a Fuzzy-PID controller using GA was intervention AUVs, in: OCEANS 2007-Europe, 2007, pp. 1–6.
[8] P. Jantapremjit, P.A. Wilson, Control and guidance for homing and docking
successfully exploited for orientational phase of docking maneuver tasks using an autonomous underwater vehicle, in: IEEE/RSJ International
of two spacecraft problem. A designed FPID controller for stabiliza- Conference on Intelligent Robots and Systems, 2007. IROS 2007, 2007, pp.
tion purpose of orientational phase of a docking maneuver is pre- 3672–3677.
[9] J. Villagra, D. Herrero-Pérez, A comparison of control techniques for robust
sented based on the SIFIMs dynamically connected PFIM for each docking maneuvers of an AGV, IEEE Trans. Control Syst. Technol. 20 (2012)
direction. The inputs of the controller are the error of Euler’s angles 1116–1123.
and angular velocities of the chaser and the driving force is set as [10] C.-C.J. Ho, N.H. McClamroch, Automatic spacecraft docking using computer
vision-based guidance and control techniques, J. Guid. Control Dyn. 16 (1993)
its output. In this work, the angular error of the spacecraft in the
281–288.
h—/—w of ‘‘Euler-axis/angle” description and the control effort in [11] H. Park, S. Di Cairano, I. Kolmanovsky, Model predictive control for spacecraft
corresponding directions are chosen as critical fitness functions rendezvous and docking with a rotating/tumbling platform and for debris
of genetic algorithm. To avoid saturation of actuators, the con- avoidance, Am. Control Conf. (ACC) 2011 (2011) 1922–1927.
[12] K. Xia, W. Huo, Robust adaptive backstepping neural networks control for
straint of maximal of the momentum produced by momentum spacecraft rendezvous and docking with uncertainties, Nonlinear Dyn. 84 (3)
wheels is added to optimization problem. By constructing this (2016) 1683–1695.
[13] N. Filipe, P. Tsiotras, Adaptive position and attitude-tracking controller for system, in: 2010 3rd International Symposium on Systems and Control in
satellite proximity operations using dual quaternions, J. Guid. Control Dyn. 38 Aeronautics and Astronautics (ISSCAA), 2010, pp. 34–39.
(2014) 566–577. [26] J. Leitner, C. Ampatzis, D. Izzo, Evolving ANNs for spacecraft rendezvous and
[14] H. Nasser, E.-H. Kiefer-Kamal, H. Hu, S. Belouettar, E. Barkanov, Active docking, in: 10th International Symposium on Artificial Intelligence, Robotics
vibration damping of composite structures using a nonlinear fuzzy and Automation in Space (i-SAIRAS), 2010, p. 8.
controller, Compos. Struct. 94 (2012) 1385–1390. [27] S.E. Mansour, G. Kember, R. Dubay, B. Robertson, Online optimization of fuzzy-
[15] L. Ping, J. Fu-Jiang, Adaptive fuzzy control for unknown nonlinear systems with PID control of a thermal process, ISA Trans. 44 (2005) 305–314.
perturbed dead-zone inputs, Acta Autom. Sin. 36 (2010) 573–579. [28] X.-G. Duan, H.-X. Li, H. Deng, Robustness of fuzzy PID controller due to its
[16] M. Marinaki, Y. Marinakis, G.E. Stavroulakis, Fuzzy control optimized by PSO inherent saturation, J. Process Control 22 (2012) 470–476.
for vibration suppression of beams, Control Eng. Practice 18 (2010) 618–629. [29] S.-K. Oh, H.-J. Jang, W. Pedrycz, Optimized fuzzy PD cascade controller: a
[17] D. Wang, X. Lin, Y. Zhang, Fuzzy logic control for a parallel hybrid hydraulic comparative analysis and design, Simul. Model. Pract. Theory 19 (2011) 181–
excavator using genetic algorithm, Autom. Constr. 20 (2011) 581–587. 195.
[18] R.-E. Precup, R.-C. David, E.M. Petriu, M.-B. Rădac, S. Preitl, J. Fodor, [30] O. Karasakal, M. Guzelkaya, I. Eksin, E. Yesil, T. Kumbasar, Online tuning of
Evolutionary optimization-based tuning of low-cost fuzzy controllers for fuzzy PID controllers via rule weighing based on normalized acceleration, Eng.
servo systems, Knowl.-Based Syst. 38 (2013) 74–84. Appl. Artif. Intell. 26 (2013) 184–197.
[19] R.-E. Precup, M.-B. Rădac, M.L. Tomescu, E.M. Petriu, S. Preitl, Stable and [31] H. Boubertakh, M. Tadjine, P.-Y. Glorennec, S. Labiod, Tuning fuzzy PD and PI
convergent iterative feedback tuning of fuzzy controllers for discrete-time controllers using reinforcement learning, ISA Trans. 49 (2010) 543–551.
SISO systems, Expert Syst. Appl. 40 (2013) 188–199. [32] M. Nie, W.W. Tan, Stable adaptive fuzzy PD plus PI controller for nonlinear
[20] S. Tong, G. Liu, Real-time simplified variable domain fuzzy control of PEM fuel uncertain systems, Fuzzy Sets Syst. 179 (2011) 1–19.
cell flow systems, Eur. J. Control 14 (2008) 223–233. [33] M. Mahmoodabadi, H. Jahanshahi, Multi-objective optimized fuzzy-PID
[21] J.N. Lygouras, P. Botsaris, J. Vourvoulakis, V. Kodogiannis, Fuzzy logic controller controllers for fourth order nonlinear systems, Eng. Sci. Technol. Int. J. 19
implementation for a solar air-conditioning system, Appl. Energy 84 (2007) (2016) 1084–1098.
1305–1318. [34] M.A. Sahib, A novel optimal PID plus second order derivative controller for AVR
[22] S. Jee, Y. Koren, Adaptive fuzzy logic controller for feed drives of a CNC system, Eng. Sci. Technol. Int. J. 18 (2015) 194–206.
machine tool, Mechatronics 14 (2004) 299–326. [35] G. Ortega, J.M. Giron-Sierra, Genetic algorithms for fuzzy control of automatic
[23] A. Sinha, S. Lyshevski, Fuzzy control with random delays using invariant cones docking with a space station, in: IEEE International Conference on
and its application to control of energy processes in microelectromechanical Evolutionary Computation, 1995, p. 157.
motion devices, Energy Convers. Manage. 46 (2005) 1305–1318. [36] J. Michael, K. Chudej, J. Pannek, Modelling and optimal control of a docking
[24] U. Zuperl, F. Cus, M. Milfelner, Fuzzy control strategy for an adaptive force maneuver with an uncontrolled satellite, arXiv preprint arXiv:1203.6782,
control in end-milling, J. Mater. Process. Technol. 164 (2005) 1472–1478. 2012.
[25] D. Zhang, S. Song, R. Pei, Dynamic obstacle avoidance of autonomous
rendezvous and docking using potential function guidance based-fuzzy logic

An Optimal Fuzzy PID Control Approach For Docking

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

An Optimal Fuzzy PID Control Approach For Docking

Uploaded by

Copyright:

Available Formats

Engineering Science and Technology, an International Journal xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Engineering Science and Technology,

Full Length Article

An optimal fuzzy PID control approach for docking maneuver of two

1. Introduction strategies have been applied to Autonomous Underwater Vehicle

Fig. 1. Schematic of designed controller structure.

Variable Value Description

JCxx 2000 Chaser angular mass around x ½kg=m2

Fig. 4. The block diagram of the optimal fuzzy-PID control.

which comprises an approximation of the ideal controller and a

formed to state space form. Here, SIFIMs dynamically connected

x_ 4 ¼ x5 Design Value Design Value Design Value

x_ 6 ¼ J1S x3 x9 J Szz J Sxx þ my K bih 0:00010 K biu 0:00495 K biw 0:00002

K rdh 0:00048 K rdu 0:00126 K rdw 0:07167

O:F:5 0:298 O:F:5 198:717

3. Results and discussion

Fig. 26. Orientation of target and chaser (dashed).

Angular Velocity (rad/s)

Fig. 27. Angular velocity of target and chaser (dashed).

Fig. 28. Momentum control of target and chaser system.

You might also like