Abstract : Thesis objectives are to develop a method to const ruct adaptive energy management strategy for regular driving patterns. The scope is a hybrid bus in a series configurat ion, with several kilomet res of elect ric driving range. The first step of this study tries to specify the optimal dist ribut ion of requested power by the driver between the two sources, the energy storage device (the bat tery) and the elect ric generator (an internal combustion engine linked to an alternator). With this aim, the optimisation problem was formalised by using simpl i fied models of the various components. Then, i t was solved by the Pontryaguine maximum principle. The resul t was integrated in the vehicle control in an optimal control mappings form, dependent on an unique parameter K. Moreover, an adaptat ion was proposed to obtain energy management strategies that can be implemented on-line. In a second step, control was adapted to develop energy management strategy for regular driving pat terns. A three level control was defined from the bat tery state of charge. The first level , that immediately results from the previous work, corresponds to the instantaneous control. Its parameterisation with K allows the local supervision (the second level) to follow a curve of desired state of charge. In the last level , this curve of desired state of charge is learnt from the a priori recognition of the followed driving pat tern. After the driving pattern execut ion, the obtained real isat ion is evaluated and possibly integrated in the knowledge base according to i ts reproducibili ty degree. Finally, the adaptive energy management strategy was validated by a Matlab/Simulink model applied to a test campaign acquired with a bus in normal operat ion conditions.