Based on the information of the bend ahead which obtained through V2X, high-definition map (HD Map), vehicle positioning or other technologies, the velocity planning and online gear decision method are explored with the actual driving state of a P2 configuration plug-in hybrid electric car when it crosses the bend, to achieve better energy economy while ensuring the driving safety. In this paper, firstly, according to the basic characteristics of the hybrid car, a simplified simulation model is built in MATLAB / Simulink to provide a verification platform for the research. Subsequently, the calculation method of safety speed in bends is established by considering the driver factor and the critical conditions when the vehicle rolls over, sideslip or oversteer. Then, according to the information of the bend ahead, the safety speed, and the general situation when vehicles cross the bend, the whole process is divided into three stages of deceleration before entering the bend, uniform velocity in the bend, and acceleration after leaving the bend. The economic velocity planning is carried out respectively in each stage and the effectiveness of the results is verified by simulation. Finally, according to the planned economic velocity, the online gear decision method based on Model Predictive Control (MPC) is proposed, and the simulation is completed in a self-defined bend scene. The results show that compared with the traditional gear shift map, the MPC gear decision method can reduce the equivalent fuel consumption by 27.64% and 22.63% separately when the initial SOC is 35% and 45%.