Self-optimizing energy management strategy for fuel-cell/ultracapacitor hybrid vehicles

Fuel-cell/ultracapacitor hybrid vehicle (FHV) needs distributing load power appropriately to its fuel cell system and ultracapacitor bank in order to minimize fuel consumption and power fluctuations in the fuel cell system while supplying adequate power to the load, and the state of charge of the ultracapacitor bank maintained at the permissible levels. This paper proposes a self-optimizing energy management strategy (EMS) for FHV to achieve this aim in an automatic way. Energy management in an FHV is formulated as the optimal tracking problem of a nonlinear discrete-time system with model bias and mixed constraints. Then, the EMS which is an artificial neural network is improved online in real time by sequentially minimizing a Hamiltonian over the driving cycle concerned. The effectiveness of the self-optimizing EMS is verified in an experimental bench, and the results are shown.