A rapid approximate dynamic programing based on constraint satisfaction for a series-parallel hybrid electric vehicle with a three-speed gearbox

Dynamic programing (DP) algorithm is frequently utilized in studies to solve the energy management schemes (EMS) of hybrid electric vehicles (HEVs). The results of this method are global optimal but computationally inefficient. This paper proposed a rapid approximate DP (RADP) algorithm to improve the time complexity to solve an off-line EMS for a series-parallel HEVs. The key step is to shorten search domain at each grid point of by calculating constraint variables and check constraints before implement Bellman equation. The results demonstrate the computing time has decreased to 1.6% of conventional DP with a slight increase of fuel consumption. And the control policy trajectories, SOC trajectories, and engine working point distribution of RADP are nearly identical to the DP in two driving missions for heavy vehicles.