Finding robust solutions against environmental changes

Many real world problems can be formalized as optimisation problems. Yet, the environment in the real world is mostly changing over time, which makes the optimisation problems change over time. This thesis is dedicated to the problem of finding robust solutions in changing environments. By robustness, we mean that a solution not only is good for the current environment but also maintains its performance after future environmental changes. The problem of finding such robust solutions repeatedly over time is termed as Robust Optimisation Over Time (ROOT). The first contribution of this thesis is a formal definition of ROOT problems and two robustness definitions of solutions to ROOT. The second contribution of this thesis is two benchmarks developed for ROOT. The third contribution of this thesis is an algorithm framework for finding solutions repeatedly over time to ROOT problems. Finally, the fourth contribution of this thesis is a dynamic handling strategy, which is used to initialise a population of solutions right after an environmental change when evolutionary algorithms are used for ROOT problems