The rapidly developing computing and communication technologies improve the autonomy of individual vehicles on the one hand and facilitate the coordination among vehicles on the other. In the context of dynamic speed management, this study considers a platoon of intelligent vehicles that are required to maintain desired inter-vehicle spaces and to respond to speed changes in a collision-free, stable and cooperative way. The platoon is modelled as a cascaded network with linear longitudinal vehicle dynamics, independent physical constraints, and coupling safety constraints. In the case of global information sharing, the authors first propose a centralised collision-free solution on the basis of model predictive control that guarantees asymptotic platoon tracking of speed changes and satisfaction of system constraints during the transient process. A cooperative distributed approach is then further proposed on the basis of the alternating direction method of multipliers resulting in a scheme involving communication only with the roadside infrastructure, e.g. the speed manager. Vehicles in a platoon conduct parallel computation while still achieving global optimal performance and coordination with respect to the collision avoidance constraints. Convergence properties of the distributed solutions are established for the concerned vehicle platoon problem. Simulation results show satisfactory platoon performance and demonstrate the effectiveness of the proposed algorithms.