Understanding the Impact of Constraints: A Rank Based Fitness Function for Evolutionary Methods

Abstract

There are design problems where some constraints may be considered objectives as in “It would be great if the solution we obtained had this characteristic.” In such problems, solutions obtained using multi-objective optimisation may help the decision maker gain insight into what is achievable without fully satisfying one of these constraints. A novel fitness function is introduced into a multi-objective population based evolutionary optimisation method, based on a plant propagation algorithm extended to multi-objective optimisation. The optimisation method is implemented and applied to the design of off-grid integrated energy systems for large scale mining operations where the aim is to use local renewable energy generation, coupled with energy storage, to eliminate the need for transporting fuel over large distances. The latter is a desired property and in this chapter is treated as a separate objective. The results presented show that the fitness function provides the desired selection pressure and, when combined with the multi-objective plant propagation algorithm, is able to find good designs that achieve the desired constraint simultaneously.