A memetic algorithm for multi-objective dynamic location problems (Q960114)

The paper considers discrete location problems with multiple time periods and multiple objectives. The new facilities may be capacitated or uncapacitated, and it is assumed that facilities may be opened, closed and reopened several times during the planning horizon. The objectives are all assumed to be linear both with respect to the location and the assignment variables. An interactive memetic algorithm is suggested for the solution of the problem. Two types of interaction with the decision maker are distinguished and integrated in the fitness evaluation used in the algorithm: The decision maker can either establish upper bounds on the (minimizing) objective functions, or reference points may be specified. Both approaches are compared on the basis of three bi-objective example problems, and the authors conclude that in these cases ``the reference point approach is capable of finding more and better solutions in less iterations than the other approach''.