Simultaneous optimisation of both scalar parameters and agentreaction strategies using genetic algorithms

摘要

When optimising simulation models of organised systems, it may benecessary to optimise the decision making behaviour of human operatorsand/or computer controllers, either because the optimal strategies forthem to use are not self-evident, or because they are dependent on othervariables which are being optimised. In this paper, a new kind ofgenetic algorithm (GA) is presented which optimises problems containingboth traditional scalar parameters and multiple reaction strategies,expressed as agent rule bases. It is a multi-level GA in the sense thatthe strings may have one or more GAs embedded within them. Itsperformance on two industrial simulation test problems indicate that itcan successfully generate good solutions to problems that haverelatively small-scale control strategies to be optimised in conjunctionwith other parameters