The energy management of Hybrid Electric Vehicles (HEV) has been the subject of a great scientific effort in recent years. Moreover, in HEV the power must be managed in real time within system constraints. The proposed approach based on a fuzzy controller, uses different set of rules depending on different phases present in a mission profile. The challenge is to compute offline these rules and to manage online a decision method to switch optimally from one rule to the other depending on the power demand. From the proposed segmentation/prediction of the requested power profile to follow, derives a switch condition between three different rules in order to decrease the fuel consumption instead of applying a unique rule computed globally on a given profile. This strategy drives the fuel cell (FC) to operate at the points of best performance. It has been verified that if this method is applied online on an unknown profile, the consumption obtained is almost optimal.