A man-machine production system (MMPS) comprising numerous operations to be
processed in a definite technological sequence, is considered. The duratio
n of each operation is random and the corresponding probability density fun
ction is pregiven. Certain operations entering the MMPS require extremely c
ostly and rare resources which have to be prepared and delivered from outsi
de and can be at the system's disposal within a short time span. Other oper
ations require non-constrained and constrained resources which are availabl
e every time period. Since the system is not fully automatic, control actio
ns, including both the resource delivering and the resource reallocation, a
re introduced by decision makers.
The problem is to predetermine the moments of delivering rare resources fro
m outside (a deterministic schedule) and to reallocate constrained availabl
e resources among operations ready to start processing (a random schedule t
o be determined in the course of the manufacturing process) in order to min
imize the system's non-operational average expenses. The problem is solved
via simulation, in combination with a cyclic coordinate descent method and
a knapsack reallocation algorithm.
The developed model has been applied on a real man-machine industrial plant
. (C) 2000 IMACS. Published by Elsevier Science B.V. All rights reserved.