Optimal Control of Triple Active Bridge Based on Deep Machine Learning Techniques

Conventionally, advanced powertrain power electronics for fuel cell (FC) powered electric vehicles may include two independent boost converters for a fuel cell stack and a battery. As the fuel cell stack and the battery may share the same DC bus and traction inverter, these power components may face certain challenges with respect to high density vehicle applications. This paper analyses the methods for designing an optimal system for integrating FC and batteries in electric vehicles and proposes a solution for optimal operation of these converters. The suggested method uses reinforcement learning to adapt to environment conditions and creates higher efficiency and consequently carbon emission reductions. This can be of higher importance for heavy duty commercial vehicles.