Hybrid platforms powered by fuel cell and supercapacitor represent a powertrain with active state-dependent constraints, providing an adverse scenario for the energy management. In these platforms, the performance of the vehicle in terms of efficiency and power compliance is noticeably affected by the energy management strategy. This paper presents a novel energy management strategy based on the estimation of the future energy demand. The strategy aims for maintaining the state of energy of the supercapacitor between two limits, which are computed online using the states of the system. The performance of the proposed strategy is tested by simulation in a hybrid electric bus operated under real urban driving conditions. The results show improvements on hydrogen consumption and on power compliance compared to the widely reported Equivalent Consumption Minimization Strategy. Also, the results include the comparison with the optimal strategy obtained offline through Dynamic Programming.