Particle swarm optimisation of a PID-based virtual coupling controller for metro railway trains

Virtual Coupling is an emerging technology that can significantly improve railway network capacity. Various advanced controllers have been proposed to achieve virtual coupling. PID-based controllers are still one of the most popular controllers for metro train Automatic Train Operation in the real world environment. It is therefore understandable that some train operators might prefer to leverage existing PID-based control systems to achieve virtual coupling. This paper focused on the parameter tuning problem of a PID-based controller by using Particle Swarm Optimisation. Such studies for virtual coupling have not been noticed in open literature. A real-world metro train and a real-world section of track were modelled. A three-train platoon using the predecessor-following topology was simulated. Nine optimisations with three different swarm sizes were carried out. The results show that, with a swarm size of 50 and maximum iteration of 1000, the optimisation was able to produce converged and consistent results. Virtual coupling simulations were carried out and analysed. The results demonstrated that the optimised PID gains are capable of achieving good precision of train spacing control, with most errors falling within the ±0.25 m range, and maximum deviations peaking around 2.6 m.