In order to study the energy-saving operation of high-speed trains, the energy consumption of trains is taken as the goal, and the speed at the transition point of the operating conditions is the optimization variable, an artificial bee colony algorithm is used to optimize the speed curve across the entire line, the purpose is to obtain the first stage optimization speed curve. On this basis, the conditions of the actual running line are fully considered, and the predictive control algorithm is used to optimize the local prediction of the speed, the purpose is to obtain the second stage optimization speed curve. The simulation results show that compared with the energy consumption in the time-saving mode, the energy consumption after the second prediction optimization is reduced by 19.29%. It is verified that the secondary speed curve obtained by the combination of the global artificial bee colony algorithm and the predictive control algorithm has better performance in energy saving effect. This paper can provide good reference value and practical significance for the energy-saving operation of other vehicles.