An evolutionary algorithm is applied to the refueling of a nuclear power plant. Refueling plans so far are designed by experts on the basis of the experience and intuition. An automatization of this process is desirable because of its high commercial and scientific interest. We develop an appropriate fitness function and parallelize the optimization process. The focal point of this paper is the comparison of two mutation operators: a naive operator, and one in which more problem specific knowledge, in particular knowledge about the symmetry of the problem, is incorporated. The latter operator reduces the search space considerably. This specialization involves the risk of excluding the best solutions from consideration. We expound a method by which to acquire some certainty that this is not the case. The method also shows how the specialized mutation operator smoothes the search space. Finally, it allows a rough estimate of the best fitness value. At this time, refueling plans found by the algorithm compare favorably with those developed by experts, but they do not yet reach the estimated optimal fitness.
[1]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[2]
Geoffrey T. Parks,et al.
Optimizing PWR Reload Core Designs
,
1992,
PPSN.
[3]
Jack Dongarra,et al.
PVM: Parallel virtual machine: a users' guide and tutorial for networked parallel computing
,
1995
.
[4]
Thomas Bäck,et al.
A Survey of Evolution Strategies
,
1991,
ICGA.
[5]
L. Darrell Whitley,et al.
The GENITOR Algorithm and Selection Pressure: Why Rank-Based Allocation of Reproductive Trials is Best
,
1989,
ICGA.
[6]
Cornelia Kappler,et al.
Evolutionary algorithms support refueling of pressurized water reactors
,
1996,
Proceedings of IEEE International Conference on Evolutionary Computation.
[7]
Thomas Bäck,et al.
An Overview of Evolutionary Algorithms for Parameter Optimization
,
1993,
Evolutionary Computation.