4th Year Project

October 2021 - May 2022

A Python-based project to optimise the pacing strategy for the Individual Time Trial event through a direct collocation method (in collaboration with British Cycling).

Previous projects had produced good results optimising the velodrome event via collocation, and it was hoped that this could be used to significantly reduce the time taken to generate power profiles for the Individual Time Trial (which was taking several hours using simulated annealing).

To implement this for my project I had to:

• Find a physiological model that was valid for longer cycling events and suitable for optimisation
• Add capability to handle courses generated from GPS data rather than mathematically-defined velodrome shapes

Overall my project was very successful - the core technical challenge of adding spline support to the collocation solver was solved within my first term, and I was free to move on to extensions like producing a power profile composed of flat segments that could be easily followed by an athlete and directly compared to the results generated by the previous solver.