An evolutionary algorithm for bilevel optimisation of Men's Team Pursuit Track Cycling

Evolutionary Computation is useful in a broad range of practical applications, however currently generalized algorithms tend to be focused upon solving problems in a theoretical domain. We aim to develop a range of generalised algorithms more suited than current algorithms to practical applications. We contextualize our algorithms using the elite sport of Team Pursuit Track Cycling, which features as part of the Summer Olympics. The sport is fiercely competitive and fractions of a second often separate the world's leading teams. We set about using Evolutionary Computation to optimise strategies for elite teams of cyclists through changes in the transition timings and the riders power outputs. We trial our range of Evolutionary Computation methods, comparing various algorithms and running them within a time frame suitable for use in a real world environment. We find significantly better results are able to be obtained through our methods than current strategies being developed at an elite level and find the use of the developed algorithms favourable for use in a practical environment.