On the relative efficacies of *cGA variants for intrinsic evolvable hardware; population, mutation, and random immigrants

摘要

Previous work has demonstrated the efficacy and feasibility of the simulated population *cGA family for embedded EH applications. This paper introduces and discusses a new *cGA variant that increases search efficacy on a specific class of EH problems without increasing the amount of hardware required for implementation. We discuss the new EA variant and provide experimental verification of its superiority against both the Dejong benchmarks and a practical problem in the construction of an EH controller to suppress thermoacoustic instability. We also compare these results with similar population-based and a population-less EAs to help understand the effects of introducing a simulated population. We conclude with a brief discussion of possible implications for the EH community in general.