An Empirical Investigation of Optimization in Dynamic Environments Using the Cellular Genetic Algorithm

Many real-world optimisation problems are dynamic. For such problems the goal is to track the progression of optimal solutions across the fluctuating fitness landscape rather than to find an exceptionally good solution for a static instance of the problem. Here we present a novel approach for creating robust solutions for non-stationary problems using the Cellular Genetic Algorithm (CGA). The CGA maps the evolving population of solutions onto a pseudo landscape. Intermediate disturbances (disasters) are introduced that break down the connectivity in the pseudo landscape, leading to isolated subpopulations. The dynamic spatial structure of the CGA helps to maintain population diversity. We investigate the performance of the algorithm using a proposed benchmark problem. Simulation results indicate that the CGA is able to respond and adapt effectively to the dynamic environment.