Case-based reasoning in scheduling: Reusing solution components

In this paper we explore the reuse of components of known good schedules in new scheduling problems. This involves accumulating a case-base of good quality schedules, retrieving a case (or cases) similar to a new scheduling problem and building a new schedule from components of the retrieved cases. We start by introducing the components of 'case-based reasoning' (CB R) and we describe a CB R solution to a 'travelling salesman problem' in order to illustrate the use of CB R in optimization problems. Two CB R solutions to a single machine scheduling problem with sequence dependent setup times are described. These are evaluated by comparing them with two more conventional alternative techniques simulated annealed and myopic search. Both CB R techniques are shown to provide good quality solutions quickly.