The Optimization Speed of Eitist Evolutionary Algorithms in Off-Line EHW

Abstract

The evolvable hardware (EHW) enables the system to be self-adaptive, self-organizing and self-repairing by incorporating evolutionary algorithm (EAs). EHW has a great application potential in weapon equipment system. However, the evolutional speed is slow which impedes its application and development and there lacks theoretical results on optimization speed of evolutionary algorithm in EHW. Mutation has been regarded as one of the key features of EAs. It is important to understand in depth the effect of the mutation operator on evolution speed of EAs. The paper gives some theoretical results by deriving the estimation upper limit of the optimization speed and the mean first hitting time of a given problem. It is shown that estimation upper limit of the optimization speed is completely determined by the transition probability and initial distribution of the population. It is also shown that the mutation probability can have a drastic impact on the optimization speed. For a given problem, the range of mutation probability can be decided quantitatively.