Beruflich Dokumente
Kultur Dokumente
*Department of Mechanical and Intelligent Systems Engineering, Pusan National University, San 30 Changieon-dong, Kumjeong-ku,
Pusan, 609-735, Korea. Tel: +82-51-510-1481, Fax: +82-51-514-0685, Email: {bj park, cdhuh }@hyowon .pusan.ac. kr
** School of Mechanical Engineering and Research Institute of Mechanical Technology, Pusan National University, San 30 Changjeon-
dong, Kumjeong-ku, Pusan, 609-735, Korea. Tel: +82-5 I-510-2454, Fax: +82-51-514-0685, Email: kshong@hyowon.pusan.ac.kr
0-7803-6562-3100510.0002000 IEEE 80
1) Nonlinear Time- Varying System Table ! Specifications of the Container Crane.
l(t)O(t) + 2i(t)O(t) + g sin(Off)) = cos(O(t))u(t) (1) Rated Load
(1 ton) Empty Load
where l(t) is the rope length (m), 0(t) is the sway angle (rad), g Speed 70 m/min 70 m/min
is the gravitational acceleration, and u(t) is acceleration of the Acceleration
Trolley time 3.0 see 3.0 see
trolley. Deceleration
2) Linear Time- Varying System time 3.0 see 3.0 see
If the sway angle O(t) is small enough, sin(0(t))~ O(t) and Speed 23 m/min 55 m/min
Acceleration
cos(O(t)) = l , then (1) can be linearized as follows: Hoist time 1.5 see 0.7 see
l(t)~(t) + 2l(t)0(t) + gO(t) = u(t) . (2) Deceleration
time 1.5 see 0.7 see
Remark: For a sway angle of + 0.1745 (rad) (well above the Therefore, the performance specification can be changed at
normal deflection of a container crane), the error in using the other acceptable level. In this paper, the control performance
linearized model is about + 0.5%. Therefore, in this paper, (2) is specifications are to maintain the traveling time of the trolley
mainly used. within one cycle and to bring the spreader to a stop within + 30
3) Linear Time-lnvariant System mm at terminal rope length.
In this model, the hoist is not considered. That is, the rope Four Stages of Operation Cycle
length is fixed at a certain value. And then, the simplest model of
the container crane is derived as follows:
l~(t) + gS(t) = u(t) . (3)
2.2. Path Planning: Four Stages of Operation Cycle
The cycle is divided into four paths as shown in Fig. 2. The
c~ c!..,...!g,!..........
four paths are described separately for the purpose of facilitating
understanding of the semi-automatic modes. In actual semi-
automatic operation mode, the four paths are continuous and at I~1 I I I I
times overlapping. (AB-movement between the safe height and the ........E..~i ...............
final landing on the trailer, BC-movement between crane's sea side A~
legs and the safe height, CD-movement between crane's sea side Chlsis Lane
legs and the target cell location, DE- movement between the clear
height above vessel and the final landing)
2.3. Specifications of the Container Crane System Fig. 2 Path planning: Four stages of operation cycle.
Specifications of the Container Crane are summarized in Table
1. In this paper, the case with rated load is considered. And, 3. Conventional Input Shaping Control
simulation parameters are as follows:
3.1. Basic Constraints for Solving Input Shaper
The traveling distance from B point to D point: 40m
1) Residual Vibration Constraints: If we assume the system is a
The rope length at A point: 30m
second-order harmonic oscillator, or can be decomposed into a set
The rope length at B point: 20m
of second-order systems, then the vibration ratio can be
The rope length at C point: 12m
determined from the expression for residual vibration amplitude
2.4. Control Performance Specification from an impulse. The vibration resulting from an input shaper is
Generally, the anti-sway system (with feedback loop) shall the superposition of the vibration resulting from the individual
bring the spreader to a stop within __. 30 mm at rope length of 25.0 impulses that compose the input shaper. The constraint on residual
meters. It shall be capable of bringing of the spreader to a stop to vibration amplitude can be expressed conveniently as the ratio of
within 2 swings or three seconds after the trolley is brought to a residual vibration amplitude with input shaping to that without
halt from full speed. However, since the input shaping control is shaping.
feedforward technique, the control action after the trolley is
V(oJ,()=e -@~" ~/(C(co,~'))2 + (S(co,())2 (4)
brought to a halt is nothing.
where,
/ C(ra, ( ) = ~, Aie (~i cos(col - ~ ' ~ t i ) (4a)
i=1
bolt
S(co, ( ) = ~ Aie(~' sin(co~/l - ( 2 t,). (4b)
i=l
rail i x 'I
motor At and ti are the amplitudes and time locations of the impulse
sequences, a9 is the system natural frequency, ( is the damping
ratio, and n is the number of impulses in the input shaper. When
l~///"f~ .......................................
V is set equal to zero, (4) is referred to as the zero vibration (ZV)
constraint.
2) Robustness Constraints: The earliest form of robust input
shaping was achieved by setting the derivative with respect to the
(xI'Yl) ~ i: ; / iii/i] TargelPo~don
frequency of the residual vibration equal to zero [2[. That is,
mg
shaper duration. This time-penalty is typical in the input shaper where t i > ti_ 1 , i = 2,3,---,n, Us(t - t i ) is equal to 1 for t > tt and
design process. 0 otherwise, and a t is a constant amplitude. The response of the
4) Time-efficiency: Time-efficiency is equivalent to time-optimality time-invariant second order linear system given by (8) is as
in the sense of satisfying all additional constraints. follows
3.3. Assumptions for Applying Input Shaping Control to
Crane Systems I n~ [. 1
Ax(t) = --~-2..,ai.~1 - - e
-
-(o~.(,-tDcos(~od(t -
}
t~) 5) . Us(t ti)
- -
83
increases with M , but the result approaches the exact solution. feasible solution. In these cases, the technique of frequency
That is, the maximum value of the sensitivity curve within the sampling can still be used to generate an approximate solution.
frequency range of interest approaches Vtot . Now, let us apply this optimization algorithm to the container
crane system depicted in Fig. 1. However, the insensitivity of the
shaper that has one hump is always wider than 0.204. Therefore, in
~. ResidualVibratic~aMay11 1 this case, it is the better way to obtain approximate solution using
+++\ ExceedDesiredLevel I I / the frequency sampling.
40It
\ I LimitVibrati°na! 1 / / Limit Vibration at Edges- ~7
~ LSpecificFrequent" I y / 3o~\'t SetVibration a,d Slope atiHumps- 17
i ~t ~re~ Vibration:~Between Edg.... dHumps-O
+ + ++
where COzk are unknown frequencies that interlace the edge and ~ 25 ...... i ...............",;. ; ........... ~............ i...... ~ + ......... ;
hump frequencies. That is, (1 - 1/2)(o < a~zl < a~hl < COz2 <.... +, 2o ........ i........... r. i ......... :.......... i ] ........ i ........ i .......
ooy
°'51 I.--~ ' ' • Optimized Shaper |
04
• 1 { ' 1 , ~ """\ ~ Rope Length Feedback / 116, pp. 654-659, 1994.
g 0.3 . J [6] Chang, P. H. and Park, J., "Use of Input Shaping Technique
0.2 ~ • •
0.1 . . . . i
with a Robust Feedback Control and Its Application to The
Position Control of the Surface Mount Machine,"
-0 1 ..... Proceedings o f the IEEE International Conference on Control
-02 Applications, pp. 397-402, 1996.
-03
[7] Pan, L. Y. and Singhose, W. E., "On the Equivalence of
Minimum Time Input Shaping with Traditional Time-Optimal
0 10 20 30 40 50 60 Control," Proceedings o f the 4th IEEE Conference on Control
time (SSC) Applications, pp. 1120-1125, 1995.
Fig. 9 Comparison of simulation results.
85