Skip to Main Content
Differential evolution has seen growing popularity as an effective yet simple evolutionary algorithm. Its main feature is that the difference between population members is used for the mutation process instead of randomly generated values. Its ease of use and implementation make it a more attractive approach to evolutionary algorithms as it is simpler to explain the choice of parameters to fit a given cost function. This paper provides a brief overview of differential evolution, and shows its uses in two applications. The first application is using differential evolution in a reference governor to generate optimal set points for the control of a power plant. The second application uses differential evolution as a gain tuning algorithm for the same power plant. Both applications include a comparison of multiple differential evolution strategies as well as a comparison with prominent particle swarm optimization techniques. Also included in this paper is a method of speeding up the convergence of differential evolution by combining it with aspects of standard evolutionary algorithms.