Page 148 - Mechatronics for Safety, Security and Dependability in a New Era

Ch27-I044963.fm Page 132 Monday, August 7, 2006 11:26 AM
Ch27-I044963.fm
132
132 Page 132 Monday, August 7,2006 11:26 AM
EXPERIMENTS

Some experiments were carried out to show: the efficiency of the approach; and the superiority of the
k-nearest-neighbor rule over the heuristics. First, we made the data on eight types of products. Next
we selected one product out of them and we made the initial labeled samples from the data on the
selected product. Then, we repeated: the selection of one product out of the remainder; the assembly
sequence planning for it by the proposed method; and the addition of the labeled samples, until all the
products were selected. We carried out the process mentioned above ten times with changing the order
of the selection of products. Figure 2(a) shows how the system inefficiency is improved through the
repeated process. In this figure, system inefficiency 57 is defined by NflNp, where Nf is the average
number of the times when a part or subassembly identified by the proposed method can not be
removed, and Np is the average number of the parts included in the given product. As shown in this
figure, the method can generate the assembly sequences more efficiently as the process is repeated.
Figure 2(b) shows the comparison of the computational time by three methods: the proposed method;
the method using the heuristics; and the method searching whole space. This figure shows that the
proposed method and the method using the heuristics considerably reduce the computational time
since these two methods reduce the search space for assembly sequences. The computational time by
the proposed method is the shortest for any of the products.
2
proposed method
• proposed method
4000
I 4000
S • method using heuristics
method using heuristics
cy ) 3000
method searching whole space
c
n 1.5 e 1.53000 • method searching whole space
e
i s
2000
c ( e 2000

i
f
f 1 m
e i 1000
n •I 1000
t
i

m 0 J I
e 0.5
t
Sys type type type type type type type type
type
type
type
type
type
type
type
type
0 0 1 2 2 3 3 4 4 5 5 6 6 7 1 1 2 2 3 3 4 5 6 6 7 8 8
7
4
5
7
types of products
Times of execution types of products
Times of execution
(b)
Computational time
(a) Transition on system inefficiency (b) Computational time
(a)
inefficiency
Transition on system
Figure 2: Result of experiments
CONCLUSIONS
We proposed an approach to the efficient planning of the assembly sequences, in which k-nearest-
neighbor rule is used to reduce the search space for the assembly sequences. We carried out some
experiments and showed the effectiveness and efficiency of the approach. We can conclude that the
proposed approach will be able to contribute much to the efficient planning of assembly sequences.
References
Bourjault, A. (1984). Contribution a une approche methodologique de Passemblage automatise:
Elaboration automatique des sequences operatories, Ph. D. dissertation, Universite de Franche-
Comte.
Lambert A. J. D. (2003). Disassembly sequencing: a survey. International Journal of Production
research 41:16, 3721-3759.
Lee S. and Yi C. (1993). Subassembly Stability and Reorientation. Proc. of IEEE Robotics and
Automation, 521-526.
Murayama T. and Oba F. (1993). An Efficient Method for Generating Assembly Sequences in Product
Design stages. Proc. of IEEE International Conference on Industrial Electronics, Control and
Instrumentation (IECON'93), 564-569.