Sunteți pe pagina 1din 11

A u t o m a m a Vol 23 N o 4 pp 437 447 1987

Prlnled

in

0005 1098 87 $ : 1 0 0 + 0 0 0
Pergamon Journals Lid
( 1987 Internatmnal Federation of Automatic Control

Great Brtlam

Time Optimal Control of Overhead Cranes


with Hoisting of the Load*
J W A U E R N I G t and H TROGER~
The time optimal control of the system which is at rest at the start pomt and has to
be at rest again at the target pomt ts determined by piecing together analytically found
solutions of the canonical system obtained from Pontryagm's maximum prmctple
Key W o r d s - - M i n i m u m t~me control, overhead crane, load homtlng, m a x i m u m principle, control
constraints, state constraints, materials handling, ship unloader

at rest again in order to be able to unload without


any time delay An additional requirement ts that
this transfer is done in minimum time Thus,
mathematically speaking a minimum time boundary value problem has to be solved
Though there exist some papers concernmg such
problems, only the problem without hoisting
motion has been solved analytically in a satisfactory
way [Manson (1977) gave the solution for a speed
controlled motor and in Auernlg (1985) this was
extended to the torque directed motor whose
behavior is determmed by its torque-speed characteristic ] Alsop et al (1965), Anselmlno and Llebhng
(1967) and Manson (1977) also considered the
hoisting motion of the load But because they did
not Include the constraint of the trolley traversmg
speed, the results are only of restricted practical
relevance
Besides the requirement that the load must be at
rest at the end of the motion, sometimes, additionally, it ts required that there is to be no swingmg
of the load in those sections of the transfer when
the trolley traverses at constant speed (Alsop et al,
1965, Mlta and Kanal, 1979) Here, however, the
authors treat the motion where only swing compensation at the target point is required Practically,
this ts the more important case and furthermore it
always results in shorter transfer times
A different type of approach has been used
in Sakawa and Shmdo (1982), where, instead of
minimizing the transfer time, the swing of the load
is minimized As the swing of the load depends
on the acceleration of the trolley, this type of
optimization will always lead to considerably
greater transfer times compared to the time optimal
case
Moreover, the type of control of the motion of
the trolley, i e posttIon control vs force control,

A b s t r a c t - - T h e m i n i m u m time transfer of a load, suspended


from a trolley by ropes, from an initial point at rest to a terminal
point where it is required to be at rest again, IS investigated by
controlling both the traversing motion of the trolley and the
hoisting m o t m n of the load Special care is given to the
modelling of the mechamcal and electrical features of the system
Mathematically, a boundary value problem w~th constraints
both m the control and state variables is gwen Necessary and
sufficient conditions for time optimal solutions are derived
using an extensmn of the Pontryagm m a x i m u m principle The
nonhnear control problem is solved analytically and a detailed
dlscussmn of the solutions is given Numerical results are
presented which apply to ship unloaders

1 INTRODUCTION

THE REQUIREMENT of faster cargo handhng work


leads to higher speeds and consequently to higher
accelerations in the motions of cranes Hence swinging of the load hanging on ropes is an unavoidable
consequence, which ts unpleasant and undesired if
it continues at the target point In this paper
attentmn will be focussed on the arrangement
shown in Fig 1 One can expect that the crane
(mamly a ship unloader or a container bridge) has
a fixed position with respect to the ship and thus
the whole unloading process is done by a planar
motion of the load hanging at the moving trolley
The basic requirement for the motion of the load
now is, that starting from point 0 in Fig 1, where
it is at rest, the load should arrive in point E being

*Received 25 February 1986, revised 11 August 1986, revised


5 January 1987 The original versmn of this paper was not
presented at any IFAC meeting This paper was recommended
for p u b h c a t m n in revised form by Associate Editor E Kremdler
under the &rectmn of E&tor H Kwakernaak
tlnstltut fur Allgemelne Maschmenlehre und Fordertechnlk,
Techmsche Umversltat Wlen, Karlsplatz 13, A-1040 Wlen,
Austria
~InstltUt fur Mechanlk, Techmsche Umversltat Wlen,
Karlsplatz 13, A-1040 Wlen, Austria
437

438

J W AUERNIGand H TROGER

III

I
J

//

M__

FIG 1 Ship unloader, showingstarting point 0 (rope length 1o)and end point E (IL) of the load, transfer distance
S of the trolley and three different ldeahzed load transfer paths A, B and C

also plays an ~mportant role m the formulatton of


the problem as will be explained below In most
papers (Manson, 1977, Mlta and Kanm, 1979,
Sakawa and Shmdo, 1982) the simpler case of
position control, i e the usage of a speed controlled
motor, has been considered or the results are vahd
only for this case (Hlppe, 1970, Kuntze, 1972)
In this paper, the control problem of the motion
of the trolley will be studied for the minimum t~me
solution of the above-mentioned boundary value
problem By means of the Pontryagm maximum
principle, solution of the minimum time control
problem is given for a trolley-load system including
the hoisting motion Furthermore, the authors
believe they have set up a realistic model, both in
its mechanical and electrical aspects As far as
possible, analytical solutions are presented, from
which the influence of the parameters can easily be
stud~ed Th~s, however, requires a restriction to
simple load paths On the other hand, in Sakawa
and Shmdo (1982) a numerical method has been
used to treat a more general path of the load, but
w~th strong conditions concerning the coupling
between hoisting and travelling motion

2 MECHANICALMODEL AND EQUATIONSOF


MOTION
A planar motion due to hoisting of the load and
traversing of the trolley for a fixed posmon of
the crane wdl be treated The vahdlty and error
estimates of the following slmphfymg assumptions
are given m Auermg (1985) The elastic defor-

/ ~ / / / ~ / / / / " / / / / / /.~//~//, /, / / / / I / /

lmL9
FIG 2 Mechamcalmodel of the load-trolley system with the
three degrees of freedom xr, qLI and the posmon Xsof the mass
center

mabthty of the crane will be neglected and tt wtll


be assumed that all elements are of mfintte sttffness
Netther dtsslpative effects hke rollmg reststance and
losses tn the drive mechantsm nor wind forces are
constdered The load ts a material point hangmg
on a massless rope The change of the rope length
due to swmgmg of the load (movement of pomt T
m Ftg 2) has been neglected
The mechamcal system shown m Fig 2 has the
three degrees of freedom xx, ~b and I For regular
working conditions and vanishing boundary conditlons the amplitudes of the swinging load and
also the steady-state angles ~b0 for constant acceleration of the trolley remain small enough such that
the equations of motion can be hnearlzed in 4~ and

Time optimal control of cranes


$o From the Langrangmn equatmns It follows,
with ( ' ) = d / d t , that
X T + m---~L (XT--I
mT

~ b + 2 I ~ b - I q~)

These equatmns are vahd both for the posmon


controlled system and the force controlled system

FT

(5)

ot = it = m L / m T

~1 T

I q ~ + 2 1 ~ b + g ~b=Xr
- - ( 1 +mE),+m_ALmH ( g + X r ~b)= Fn-mH ,1,
where m L is the mass of the load, and mT is a
reduced mass includmg the mass of the trolley and
the equivalent contributions of the rotating parts
of the trolley travel mechamsm Furthermore, w~th
the radius r of the rope drum the reduced quantities
m . = l n / r 2 and F n = M H / r of the hoisting mechanIsm have been mtroduced
There are two basically different cases for the
control of the trolley traversing motion Either
xr = at(t), which means that the position of the
trolley is controlled (position controlled system), or
F r = FT(t), which is a force controlled system The
following control variables are mtroduced
Fr(t)

aT(t)
U T

439

--

or

UT

aTm,,t

--

can be used as a parameter to &stingulsh these two


cases in (4) The posmon controlled system has only
two degrees of freedom (~ and 2) because a is
prescribed Hence there is no coupling between a
and * For posmon controlled systems
= 0

must be set in (4), whereas for force controlled


systems equation (5) is vahd
In the practical operation of cranes, 2" < 0 2 << 1
is valid (Schlemmlnger, 1970) Additionally, the
load hoisting acceleration time is only a small
fractmn of the whole trolley traversing time Hence
the influence of load hoisting acceleration on the
swinging of the load is negligible and the assumption of a piecewme constant hoisting speed Vn ~s
well justified With the dimensionless hoisting speed

(2)

VH

fTm,,

VH-- -

for the position controlled system or the force


controlled system, respectwely aTma, m (2) for the
position controlled system ~s given by

(6)

x/g lmm
the equations of motmn (4) now read
O"' "4- ~ (1) m~- UT

aTm.. = max (laT(t)l)

2 ~"+2

to<--t<--t E

2' O ' + ( 1 + ~ ) ~ = U T

and for the force controlled system


fTma x
aTmax-

FTmax=

mT

max

to<-t<-tL

~' __ VH

It can be seen from the second part of (7) that for


lncreasmg ~ the period of the oscillation of the load
decreases Hence for the force controlled system a
faster swing compensation is possible

(IFT(t)I)

Introduction dimensionless quantities

T = fDma x t,

o" - -

('Dmax= / ~ m ~ n '

XT g
-

aTmax /ram'
(I)=

q~ and

aTmax

l
2 = 7-,
lmm

(3)

with ( )' = d/dr, the equations of motion (1) read


e"+cz (1--2") q)=u T
2 *"+2
--(1+

2' q ) ' + ( 1 + ~ (1--2")) O = u T


m~nL
) ) , , + mL
mn

__Fn
mH g

(7)

(4)

3 MODEL OF THE TROLLEY DRIVE MOTOR


CHARACTERISTICS
The quality of approximation of the real behavior
of the trolley drive motor has a decisive influence
on the practical relevance of the results to be
obtained Basically there exist two different cases
The first case is that of a speed controlled motor
whose rotational speed is determined by a feedback
control loop The related characteristic quantity is
at(t), which is assumed to be a plecewlse constant
aT(t)=ar,,

t,_ l < t < t,

(8)

This simplest form of a time-dependent function


allows us to construct the solution of the boundary
value problem by piecing together the analytical
solutions of (7) Furthermore, it will be seen that
this assumption is adequate to describe the time

440

J W AUERNIG and H TROGFR

:",

1L---.

controlled and p o s m o n controlled sectmns

"LJ-_-&
M

-I?

"'"II

II OV,..~

-4

....

_L--~

'

UT =

{OT

If O" :j~ ~'T ....


If ~' = VTm..

(12)

Both real ship unloaders and most real container


cranes (especmlly those e q m p p e d with a spreader
rotating device) satisfy /~ < 1 (Auernlg, 1985,
Franke, 1973) Thus In these important applications
the undesirable n o n - c o n t i n u o u s m o t o r characteristic can be replaced by force controlhng and the
velocity constraint (9)

r, M
r,

{0

4 FORMULATION

91....

FIG 3 Motor characteristics (1) &rect-current shunt-wound


motor, (2) three-phase sqmrrel-cage motor, (3) Anselmlno and
Ltebhng (1967), (4) Hippe 0970) and Kuntze (1972), (5) present
paper

OF THE OPTIMAL CONTROL


PROBLEM

To simplify (7), new variables a s ( p o s m o n of the


center of mass, Fig 2) and fl are defined by
~X

a s = a - ! -+ ~
optimal solutmn
constraint

A d d m o n a l l y to (8), the speed

fl=/

(13)

,Xs) = (Crs, a's,fl, fl',/)

(14)

W~th the state vector


XT = ( X 1 , X 2 ,

}XTI ~ VTmax

and

(9)
and the control vector

must hold The speed controlled m o t o r yields


p o s m o n control of the t r o l l e y - l o a d system
In the second case the natural b e h a w o r of a m o t o r
~s characterized by its t o r q u e - s p e e d characteristic
Such a m o t o r will be called "torque &rected" The
authors obtained

aT = ( 1' U2 ) = (UT' VH )'

(15)

the following set of n o n h n e a r first order ordinary


dtfferentml e q u a t m n s is obtained
X"1 ~ X 2

F T = FT(XT(t), t)

(10)
1+~

as a related quantity The t o r q u e - s p e e d characterlStlCS of a real &rect-current s h u n t - w o u n d m o t o r


and of a real three-phase sqmrrel-cage motor,
together w~th some too academm charactensttcs
other authors have used and the ldeahzed characteristic which the authors have used here, are
c o m p a r e d in Fig 3 The authors' ldeahzed t o r q u e speed c h a r a c t e n s u c (case 5 m Ftg 3) is fixed if the
m a x i m u m starting torque is prescribed a s Msm.x,
the no-load speed as n o and the m o t o r control
varmble as ti(t), whmh is assumed to be pmcewlse
constant again
liT(t) = liT,,

t,_ t ~ t ~ t,

(1 1)

FTm,, and VT...x are the corresponding quantities for


the translatory m o t m n
The investigation of the behavior of the trolleyload system shows that for values ~ _< ~max ~ 1 the
motion of the trolley can be dwlded into force

xk = - ( 1 + ~)

X3 +

Ul
(16)

X'5 = U2

The state variables must satisfy the b o u n d a r y


condmons
X(Zo)w = (0,0,0,0,20)

(17)

X(TE)T = (Z, 0,0,0,2E)

(18)

where X is the dimensionless distance S (Fig 1)


according to the third part of (3) The control
variables are b o u n d e d by
--1 _<u 1_< 1

and

- v n . , a x < u z <_ vn..." (19)

Practically, two important state constraints follow


for the traversmg speed of the trolley [from (9) and

Time optimal control of cranes


the first part of (13)] and for the rope length. They
are

g l = VTma x

--

X2

--

1+~

x4 > 0

and

g 2 = x 5 - 1 >_0

(20)

respectwely, where

Gf

VTmax

~lm,.

aTmal

441

pracUcal unloading work, and (B) and (C) are


limiting cases The hmltmg case (C) (constant hoistlng speed d u n n g traversing of the trolley) will now
be considered, In order to determine the minimum
hoisting speed The trolley traversing section of
case (B) is further a llmtlng case of (C)
There is no control of the hoisting motion The
hoisting speed and the rope length are determined
by the fifth boundary conditions (17) and (18), for
which the following are set
20 = lo/l E >_ 1,

As a minimum time solution is required, the optimality c o n d m o n is given by

f?

2'

1 - ;to
TE

dt ~ max

(27)

This gives

2(~) = 20 + 2' T,
-

2E = 1

--

(28)

T O

(21)
The Hamlltonlan H IS given by

To solve problem (16)-(21) use is made of the


Pontryagln m a x i m u m principle Necessary and
sufficient conditions for the optimal solution of the
problem

fto

EfO(x(t),u(t),t) dt

---1.

max

(22)

H = - 1 + Pl x2 + P3 x4 - (1 + ~) P4 x3
2

+(Pf~+~+Pa) ul

(29)

With the velocity constraint given by the first part


of (20), this gives the Lagrangtan

subject to
x(t) = f(x(t),u(t),t),x,f~R~,u~R ",
X(to) = x0

and

X(tE) =

u(t) e U ~ R',

xE,

(23)
(24)
(25)

L = H + ql

I"Tmax - -

x2 - -1 +

x,

(30)

The adjoint equations for P read

p]=O

p~=(l+~)P~
2

g(x(t), t) > 0,

g e R~

P2 = --Pl + ql
can be found In Kamlen and Schwartz (1981) on
pp 220-221 and in Theorem 8 in Selerstad and
Sydsaeter (1977), where in addition to the usual
necessary conditions for optimality also sufficient
conditions making use of the concavity of the
maximized Hamlltonlan and the state constramts
are formulated
5 RESULTS

Though the state equations (16) and the corresponding adjomt equations are nonlinear and
coupled, analytic solutions can be given This is
done by obtaining solutions of the unconstrained
and the constrained problem and piecing them
together

5 1 Constant hmsttng speed


In F i g 1 three qualitatively different cases [ ( A ) (C)] are shown (A) is the case regularly found in
&OT

23

4-8

(31)

(26)
P4 = - - P 3 + T - ' 7 - - 1W~

ql

Since H is linear In ul, its maximum is obtained


for
+1

udO =

e[--1,
--1

>0
+1]

If P2(Z) + p,(T)

=0
<0
(32)

The first four state equations (16) and the adjolnt


equations (31) constitute a canonical problem of
order eight The initial values p(To) must be selected
in such a way that the first four boundary conditions
(18) are satisfied However, it can be seen that the
x, and the p, are related to each other only by the
rope length 2 Thus with the general solution of
(3 I) and the control law (32) the qualitative behavior

442

J W AUERNIGand H TROGER

of the control is given If there exists a solutmn x(r)


for th~s control, then the necessary condlUons for
opt]mahty are satisfied Furthermore, H is hnear
m x and u Thus the maximized Hamdtoman ~s
concave Finally, also the first part of (20) is hnear
m x Thus both suificlent condmons for opt]mahty
are satisfied
Now results for two different cases are gwen
51 I
g~ > 0,
follows
be zero

Unconstrained trolley traversing speed If


]e the velocity constraint is mactwe, ~t
from the maximum prmople that q~ must
Thus the general solutmn of (31) reads

p2

!,+0<

- 1:)4

T02

T23

T~Tss

-1

~-"

_%/

t~

p] = A
P2 = B - - A
P3 --

--2 (1 + ~t)
,~,
[C Jo(z) + D Y0(z)]

T
[

pa = z [C Jr(z) + D Yt(z)]

Toe

(33)

with
2
z= ~

~/(1 + ~) )

(34)

where d0, J r , Yo and Y1 are Bessel Functtons of


the first and the second kind of order 0 and l
According to (32) and (33), the opUmal control is
determmed by the s~gn of the sum of a hnear and
an oscillating functmn For small d~stances Y. the
most general optimal solution has three switching
points as shown m F~g 4 T h s bang-bang solutmn
is stmllar to the well known symmetric solutmn for
constant rope length (for instance see Htppe, 1970,
values of Z, h m m n g cases with one sw]tchmg point
also extst
Instead of solwng the canomcal system of the
first four state equations (16) and the adjomt
equations (31) by determining the mmal values P(~o)
or the integration constants A, B, C, D m (33)
respectwely, the time intervals Zo2, T23, "t'3s and
"L'56 are introduced as new unknowns They are
determined by the first four end condmons (18)
The reqmred values X(ZE) are gtven by the solutmn
U1

x~ = a + b

x2 = b +

u]
l+ct

U1
"(3-- - -

l+ct

x, -

Ul

T2

r + l+~

1 + ~

"t"

2 + z [c J l ( z ) + d Yl(z)]

2,+2

(1 +or)
2'

[ c Jo(Z) + d

Yo(z)]
(35)

of the first four state equations (16), the first four

FIG 4 Adjomt variables P2 and P4, optimal control u~ 132),


speed x2 of the center of mass and a' of the trolley, swing angle
q~of the load and rope length / for constant hoisting speed and
unconstrained traversing speed

mmal values (17) and the reqmrement that the state


varmbles m the three switching points z2, za and
~4 must be continuous For larger distances ~z
several solutmns can occur Cons~denng first the
simple case of constant rope length it can eastly be
understood that also T23 + n ~ and za5 + n
boundary conditions (18) In order to get the fastest
solution, the solution with the smallest values of
these two time mtervals must be selected
The constants A, B, C and D are now obtained
from the swltchmg condition (32) at the switching
points z2, z3 and ~4 and the condition that the
Hamdtontan becomes zero at the free end pomt rE
(Kamlen and Schwartz, 1981, pp 143-150) The
solutton ts ttme optimal, if no further swltchmg
points exist This is always the case
Figure 5 shows, for increasing traversmg dstance
E, the vanatmn of the locatmn of the switching
points L(E) and of the corresponding time intervals
L~(Y.), together wtth some typical relatlonshtps of
the velocity x2 of the center of mass
5 1 2 Constramed trolley traversing speed For
short trolley traversing distances (Z small), case
5 1 1 always holds because the trolley does not
reach the maximum traversmg speed VT=.. HOWever, if 5". is big enough such that VT=.. tS reached,
the constramt gwen by the first part of (20) is
actwated, resulting m gl = 0 Differentiating the

443

Time optimal control of cranes

2s

~-0
X,-2

-t

50
O0

,/

-- 1 B

l-

-:-//,/----~
/ / " , , '

i It

'

"1

1~

~ -

20

30

40

'

tl t

50

50

70

80

'

'

90

l
~O0

'

'

ll0

120

---.-Z
F~G 5 Dependance of the traversing time Zo6, the switching points L and the corresponding ttme intervals %
on the traversing distance Z (see also Fig 4)

first part of (20) with respect to z and introducing


x2 and x, from (16) gives

X3
/.

ul = g - -

(36)

In the case of an actwe constraint g~, the general


solutmn of (16) reads

xl = a '{- VTmax ~"

1+

X3

~x

X2 = VTm'x- - 1- +If ul saUsfies the first part of (19), whmh always


holds according to our chome of ~, there is a
singular solution to the control problem, ~e the
switching funcUon gwen by the second part of (32)
is zero Dlfferentmtmg the second part of (32)
and introducing (31) gwes the multlpher function
qt = Pt/( 1 + ~) + P3, which allows one to find the
general solution of (31) as

pl=,4
P2 = - - ( 1 + ~X) P4

P3--

- - 2 ( 1 +~x) [C Jo(z)+O
2'

Yo(z)]

cx

+ 1--~

Pl

P4 = 2 [(~ J l ( z ) + / )

YI(Z)]

(37)

with

2
z = i-~?1 ,,//~"

(38)

x3 =

[6 Jl(z)

x4
+ i[

YI(z)]

x4=-~ [6 Jo(z) + 7a yo(z)]

(39)

To obtain the optimal solution of the motion of


the trolley one must combine the two solutions (33)
and (37) for the adjomt variables and the solutions
(35) and (39) for the state variables under the
c o n d m o n that the state varmbles must be continuous The adjomt variables Pl and P3 are continuous,
P2 and P4 could have jumps However, it turned
out that for the problem at hand P2 and P4 are
found to be continuous too
Starting from the system w~thout velocity constramt (Fig 4), the system behavior with velocity
constraint as presented in Fig 6 will be explained
As long as a' at zl or z4 is smaller than VTmax, the
constraint gl ~s inactive and the control shown m
F~g 4 exists However, ff in "c1 or m z4, a' = VTmax
~s reached, an additional sectmn of constant velocity
motion must be included Its t~me period can be
calculated from the c o n d m o n that a'(z) = vr... for
z = zt or r = z4, respectively Depending on the
system parameters this period can occur before z2
or z5 or z2 and zs

444

J W AUERNIG a n d H TROGER
p2

- ~.4

llx ~ 2 3 _

__

Section 5 1 2 does not result m a time o p t i m a l


solution a n d as m & c a t e d at the end of th~s section
a new strategy w~th m o r e c o n t r o l intervals ~s
necessary

- ~ -

5 2 Ttme opttmal traversmg and hmstmg motton


Th~s p r o b l e m has not yet been treated analytically m the literature, obviously, because it poses
great m a t h e m a t i c a l difficulties In o r d e r to understand the b e h a v i o r of the system q u a h t a t w e l y the
a u t h o r s conmder the special case where

X2~ (7'

/o
l+~

/E,

(41)

'T'
I

FIG 6 As m Fig 4 but for constrained trolley traversing speed

TABLE 1 OPERATING (YCLES OF SEVERAL SHIP UNLOADERS,


EQUIPPEDWITH SPEEDCONTROLLEDMOTORS(0~ = 0)

Ship unloader
A

VK..,
4 18

4 40

321

D
E

461
3 04

ZO
278
2 06
1 94
2 29
1 13
1 76

Z
3597
34 31
1776
22 44
2998
23 08

E*
3602
34 63
2437
25 62
3013
22 39

l e the l o a d ~s at the s a m e height at the beginning


a n d at the end p o i n t of the m o t i o n Similarly as m
Section 5 1, one can write d o w n H, L a n d the system
of the a d j o i n t e q u a t i o n s which is of fifth o r d e r n o w
In a s t r a i g h t f o r w a r d calculation H becomes a
m a x i m u m if one selects ut a n d U2 as
+1
ul(O =

~[+1,

>0
-1]

ff

- - P2
+p+
l+ct

=0

-l

<0
(42)

and
The b e h a v i o r s h o w n m F i g 6 is vahd as long as
the a d j o m t variables satisfy

p2(r-~) +p+(z)(~O

z<--z<zl

1+~

rs<z_<z

_0

+vH ,x
u2(r) =

(40)

[ - - VHrnax,

+ VHm.x]

-- VHmax

If

P5

0
0

(43)
E q u a h t y m (40) defines the m a x i m u m & s t a n c e
Y = Z* where the c o n t r o l strategy s h o w n in F i g 6
~s the t~me o p t i m a l strategy F o r a further increase
of X, q u a h t a t w e changes m the c o n t r o l variable ul
occur T h e n m that p o i n t where (40) is zero,
an a d d m o n a l d e c e l e r a t i o n o r acceleration interval
must be inserted

5 1 3 Apphcatwn to ship unloaders A n a p p h c a t i o n of o u r theoretical results is the e v a l u a t i o n of


c o n t r o l strategies of ship u n l o a d e r s Here it is
checked w h e t h e r the o p t i m a l strategies f o u n d are
a p p l i c a b l e This is d e o d e d by checking w h e t h e r
the transfer distances Y. of practically o c c u r r i n g
w o r k i n g cycles of existing ship u n l o a d e r s are smaller
t h a n E* as defined by the e q u a l i t y m (40) T h e
a u t h o r s c o n s i d e r e d five &fferent ship u n l o a d e r s
o p e r a t i n g m E u r o p e a n p o r t s whose d a t a are given
m A u e r n i g (1985) T h e results are given m T a b l e 1
a n d it can be seen that in the four cases [ ( A ) - ( D ) ]
y. < 5"+* holds In the fifth case (E), 5". is shghtly
bigger t h a n E* In this case the strategy given in

N o w the two systems of differential e q u a t i o n s for


the state variables a n d the a d j o l n t variables are
s t r o n g l y c o u p l e d F o r s t u d y i n g the b e h a v i o r of the
s o l u t i o n of the c a n o m c a l system of tenth order,
the a u t h o r s a d d i t i o n a l l y assume that there IS no
restriction c o n c e r n i n g the traversing velocity of the
trolley a n d that the c o n s t r a i n t c o n c e r n i n g the rope
length can be t a k e n care of later
T h e initial values P(Zo) have to be selected m
such a w a y that the b o u n d a r y c o n d i t i o n s (18) for
the state variables are satisfied F u r t h e r m o r e , ~t is
i m p o r t a n t to note t h a t the solutions of the first four
state e q u a t i o n s (16) a n d the first four a d l o m t
e q u a t i o n s are identical to those c a l c u l a t e d before
But z' = u2 a n d 2 = x5 m u s t be substituted in (33)(35) The r e m a i n i n g state e q u a t i o n given by the
fifth p a r t of (16) a n d the fifth a d j o l n t e q u a t i o n can
be solved by q u a d r a t u r e ( A b r a m o w l t z a n d Stegun,
1984, p 177) in the form
x s = e + u2 T

(44)

Ttme optimal control of cranes

445

'l

UH.o=

X=~0"

FIG 7 O p t i m a l control ut of the t r a v e r s i n g m o t i o n a n d u 2 of the hoisting m o t i o n for the s a m e load height at


the s t a r t i n g a n d end point a n d the p a r a m e t e r values ct = 0 5, ,to = 2, Z = 15, vtt.., = 0 2 The n u m b e r s (1)-(4)
c o r r e s p o n d to the transfer times toe given m T a b l e 2

TABLE 2

TRANSFER TIME ZOE CORRESPONDING TO THE IDEALIZED


LOAD PATHS IN FIG 7

Case

vH..,

toe

1
2
3
4

02
01
0
-01

967
997
10 26
10 54

u:%

-~"-'- t

7"

and

Ps = E +

u!

[C Jo(z) + D YI(Z)]

ko

U2

x,,,V

4 (1 + ~t)2
+
u3
{[C Jo(z) + O Yo(z)]

[c Jo(z) + d Yo(z)]
+ [C Jl(z) + D Yl(z)]
[c J,(z) + d Y,(z)]}

..... ~ ......

~2

FIG 8 As Fig 7, but for the p a r a m e t e r values c = 0 5, z o = 2,


= 50, v . . . = 0 1 The n u m b e r s (1)-(4) c o r r e s p o n d to T a b l e 3

(45)

F r o m the control law (42) and the solution for


the adlomt variables it can be seen that we have
the same situation as in part 5 1 Hence the optimal
solutton for the traversing motton of the trolley ~s
slmdar to that shown m Fig 4
Startmg from the ltmltmg case VH=.x= 0, the time
opUmal control of the traversing and the hoisting
motion can be obtained Three typtcal examples
are gtven m Figs 7 - 9 In the dmgrams for xs also
other ldeahzed subopttmal rope length relatlonshtps
are shown The corresponding traversmg ttmes for
Figs 7 - 9 are given m Tables 2 - 4
In Ftg 7 and Table 2 the solution for short
distances Z is presented That the load is hfted m
the first half of the motton and lowered m the
second is physically plausible because for decreasing
rope length the period of osctllatlon of the load
also decreases It is obvious that with increasing
hoisting velocity the traversing time is reduced
Increasing the d~stance ~ results m a solution as
shown m Fig 8 By means of a comparison with
subopUmal control strategtes indicated by the
different rope length relationships m Table 3 tt is

TABLE 3 TRANSFER TIME "t'OE CORRESPONDING TO THE IDEALIZED


LOAD PATHS IN FIG 8

Case

roe

1
2
3
4

17 39
1748
17 57
17 82

shown that the found strategy Is optimal Increasing


the hoisting speed m comparison to F~g 8 results
m Fig 9, from which it can be seen that two
additional switching intervals occur In Table 4
suboptimal strategies are compared w~th the optimal again
Because the Hamlltontan IS nonlinear m x, the
proof of concawty of the max~mtzed H a m d t o m a n
could not be gwen If several solutions exist satlsfymg the necessary condmons the fastest one of them
must be picked
6 CONCLUSIONS

Whereas the mechamcal modelling of the system


with three degrees of freedom ~s qmte straightforward, special care Is given to the modelhng of the
electrical drive of the trolley Here it is possible to

446

J W AUERNIG a n d H TROGER

1"

'

X~s x o
. . . . . .

~'22--

........

FJG 9 As F~g 7, but for the parameter values :t = 0 5, z o = 2,


I: = 50, VH.,, = 0 15 The numbers (1)-(4) correspond to Table 4

TABLE 4

T R A N S F E R TIME

TOE C O R R E S P O N D I N G

TO T H E I D E A L I Z E D

LOAD PATHS IN FIG 9


Case

1
2
3
4

TOE

17 329
17 331
17 498
17 568

include m one m a t h e m a t i c a l d e s c n p t t o n b o t h cases


of m o t o r s used m o p e r a t i n g systems, l e the speed
c o n t r o l l e d a n d the t o r q u e directed m o t o r , where m
the m o d e l used the f o r m e r is a specml case of the
latter
In the m a t h e m a t i c a l formulatson of the o p t i m u m
c o n t r o l p r o b l e m an extension of the P o n t r y a g m
m a x i m u m principle is used which allows one to
o b t a i n necessary a n d sufficient c o n d m o n s for optlm a h t y T h o u g h the state e q u a t i o n s a n d the a d j o m t
e q u a t i o n s are c o u p l e d the a u t h o r s were able to give
a n a l y t i c solutzons of the c a n o m c a l e q u a t i o n s m
the c o n s i d e r e d case of ptecewlse c o n s t a n t hoisting
speed The o p t t m a l s o l u t i o n basically is a b a n g b a n g c o n t r o l which ~s d e t e r m i n e d by a switching
function c o n t a i n i n g Bessel functions
By m e a n s of the s o l u t i o n of the specml case of
c o n s t a n t hoisting speed d u r i n g trolley t r a v e r s m g
[(C) m F i g 1], the a u t h o r s were able to e v a l u a t e
practically used o t h e r c o n t r o l strategies especmlly
for ship u n l o a d e r systems T h e s o l u t i o n gwen
by M a n s o n (1977) is e x p a n d e d to systems wtth
c o n s t r a i n e d t r a v e r s i n g speed a n d systems driven
by t o r q u e directed m o t o r s T h e H a m t l t o m a n is
concave a n d hence all the necessary a n d sutficlent
condtt~ons for o p t t m a h t y are sat]stied T h e m o s t
~mportant p a r a m e t e r s m all the a u t h o r s ' cons~derattons are the transfer d~stance Z a n d the m a x i m u m
speed vr.,.x of the trolley, which t o g e t h e r wtth the
osctllatlon b e h a v t o r of the l o a d s t r o n g l y influence
the c o n t r o l strategy G e n e r a l l y speaking, one can
say that the s h o r t e r the p e r i o d of o s c d l a t t o n of the
l o a d the faster the t r a n s e r can be m a d e T h u s the

t o r q u e dtrected m o t o r is s u p e r i o r to the speed


c o n t r o l l e d m o t o r The longer the transfer distance
E of the trolley, the m o r e c o m p l i c a t e d the c o n t r o l
strategy will be However, the a u t h o r s have shown
for cranes o p e r a t i n g m E u r o p e a n ports, that the
strategies whtch they have designed a l m o s t always
apply
F u r t h e r , they also treated the time o p t i m a l control of the hoisting m o t i o n , however, only for the
simple case where the two end p o m t s of the transfer
are at the same height In this case only the
necessary c o n d i t i o n s of the o p t l m a h t y t h e o r e m are
satisfied a n d the a u t h o r s had to pick out the fastest
s o l u t i o n from the set of admtsstble solutions
F m a l l y , the a u t h o r s beheve it to be necessary to
c o m m e n t shortly on a second i m p o r t a n t aspect of
the o p t t m l z a t t o n of such an u n l o a d i n g system, J e
the energy c o n s u m p t i o n The a u t h o r s d i d not
consider this aspect at all, however, they w a n t to
r e m a r k that the time o p t i m a l c o n t r o l of the h o m t m g
m o t i o n results m a higher energy c o n s u m p t i o n As
for practically o p e r a t i n g systems, the rating of
hoisting m o t o r s is a b o u t three to five times that for
traversing m o t o r s , m the next step of o p t i m i z a t i o n
the energy c o n s u m p t i o n of the hoisting m o t i o n
should be included
Ackno~ledgements--Th~s work was partly supported by the

Hochschul.lubdaumsstlftung der Stadt Wren and by a grant of


the Fonds zur Forderung der Wlssenschafthchen Forschung m
Austria under project P5519
REFERENCES
Abramowltz, M and I A Stegun (1984) Pocketbook of Mathematlcal Functions Abridged edition of Handbook of Mathematwal Functions Harrl Deutsch, Thun- Frankfurt am Main,
FRG
Alsop, C F, G A Forster and F R Holmes (1965) Ore
unloader automatlon--a feasibility study Proc 1965 IFAC
Syrup, 295-305
Anselmlno, E and T M Llebhng(1967) ZeltOptlmale Regelung
der Bewegung elner hangenden Last zwmchen zwel beheblgen
Randpunkten Proc 1967 Int Analogue Computation Meetings, i, 482-492
Auermg, J W (1985) Steuerstrateglen fur Laufkatzkrane zur
Vermeldung des Lastpendelns im Zlelpunkt Dissertation,
Techn Umv Wlen, Austria
Franke, R (1973) Ober die Unterdruckung des Lastpendelns
be1 Laufkatzkranen, insbesondere von Contamerkranen Dissertation, Techn Hochschule Munchen, F R G
Hlppe, P (1970) Zeltopumale Steuerung emes Erzentladers
Regelungstechmk und Prozefl-Datenverarbettung, 18, 346350
Kamlen, M I and N L Schwartz(1981) DynamwOptlmlzatlon
Elsevier North Holland, Amsterdam, New York
Kuntze, H -B (1972) Bletrag zur zeltopumalen Steuerung und
Regelung spezleller schwmgungsf'ahlgerSysteme (zum Betsplel
yon Laufkranen) Dissertation, Hochschule fur Verkehrswesen, Dresden, G D R
Manson, G A (1977) Time optimal control methods arising
from the study of overhead cranes Ph D Thesis, Umv of
Strathclyde, Glasgow, U K
Mlta, T and T Kanal (1979) Optimal control of the crane
system using the maximum speed of the trolley (m Japanese
with Enghsh abstract) Trans Soc Instrum Control Eng
(Japan), 15, 833-838
Sakawa, Y and Y Shmdo (1982) Optimal control of container

Time optimal control of cranes


cranes Automatlca, 18, 257-266
Schlemmmger, K (1970) Angabe des Gutdurchsatzes von
Schlffsentladern mR Grelferbetneb Hebezeuge F6rderm~ttel,
10, 321-325

447

Selerstad, A and K Sydsaeter (1977) Sufficient conditions m


optimal control theory lnt Econ Rev, 18, 367-391

S-ar putea să vă placă și