Hybrid generation combines advantages from fuel cell systems with non-predictable generation approaches, such as photovoltaic and wind generators. In such hybrid systems, it is desirable to minimize as much as possible the fuel consumption, for the sake of reducing costs and increasing the system autonomy. This paper proposes an optimization algorithm, referred to as population-based incremental learning, in order to maximize the produced power of a photovoltaic generator. This maximization reduces the fuel consumption in the hybrid aggregation. Moreover, the algorithm's speed enables the real-time computation of the best configuration for the photovoltaic system, which also optimizes the fuel consumption in the complementary fuel cell system. Finally, a system experimental validation is presented considering 6 photovoltaic modules and a NEXA 1.2KW fuel cell. Such a validation demonstrates the effectiveness of the proposed algorithm to reduce the hydrogen consumption in these hybrid systems.
fuel cell, reconfiguration of photovoltaic systems, hybrid generation, population-based incremental learning.