An iterative dynamic programming approach for the global optimal control of hybrid electric vehicles under real-time constraints

Dynamic programming is a well-known method to evaluate and generate the optimal control of hybrid electric vehicles (HEV) on given driving cycles. Because of its computational complexity the method per se is not suitable for realtime implementation especially with higher dimensions. In this paper an iterative dynamic programming (IDP) approach is proposed to reduce computing time by converging the optimal control strategy iteratively within an adaptive multidimensional search space. The velocity is investigated as a new degree of freedom to include the influence of a variable driving trajectory. The approach is implemented as a nonlinear model predictive controller, with the main focus on savings in computing time.