This paper generalizes a redundancy optimization problem to multi-stat
e systems, where the system and its components have a range of perform
ance levels - from perfect functioning to complete failure. The compon
ents are: 1) chosen from a list of products available in the market, a
nd 2) characterized by their nominal performance level, availability a
nd cost. System availability is represented by a multi-state availabil
ity function, which extends the binary-state availability. To satisfy
the required multi-state system availability the redundancy for each c
omponent can be used. A procedure which determines the minimal-cost se
ries-parallel system structure subject to a multi-state availability c
onstraint is proposed. A fast procedure is developed, based on univers
al generating function, to evaluate the multi-state system availabilit
y. Two important types of systems are considered, and special operator
s for the universal generating function determination are introduced.
A genetic algorithm is used as an optimization technique. Examples are
given.