It is worthy to say that the Electric Vehicle (EV) with their batteries are a cornerstone to be positioned in the new automotive market industry. Platforms such as Batteries Europe are a great example. Nowadays, Lithium-ion batteries are the ones used for this technology because of its properties. Nevertheless, this raises new safety and reliability challenges which require development of novel sophisticated Battery Management Systems (BMS) which control and monitor the battery system. However, current BMS does not fit well automotive requirements, and improvements are demanded. And the strategic decision of that is clear: better BMSs will produce benefits such as less battery degradation, better performance and more lifetime. The key to improve the BMS is to employ more complex battery models on-board. However, they are highly-time consuming to be used. Therefore, we propose a methodology to deal with the above problem, achieving great results. We developed a technique able to learn a highly accurate physic-based model (Newman’s P2D Model), deleting the elevated number of degrees of freedom and dimensionality of the original problem. In addition, an excellent agreement was observed. Furthermore, due to the low computational cost of the created model, it can be perfectly integrated on-board of the EV as well as in system simulation tools such as SimulationX. Therefore, we integrated the proposed battery approach in SimulationX to simulate the whole EV system. We would like to note that there is a big advancement. The accurate battery models such as the Newman’s P2D Model cannot be integrated in this type of tools because they are highly-time consuming. However, since we are able to achieve a reduction in computational cost that is thousands of times lower (maintaining good accuracy), we have no problem in using the discussed approach for these applications. Thanks to that, we developed an innovative planning algorithm in SimulationX to make decisions based on predictions of the whole EV, taking into consideration this fast and accurate battery model. For example, we can decide the best possible itinerary considering different battery criteria (where several itineraries are quickly simulated to select the best one). Or another example, the algorithm is also capable of proposing changes in the driving behavior if an itinerary is maintained but it detects that battery problems can arise.
Mr. Abel Sancarlos, ESI Group/ENSAM/UZ, FRANCE Ms. Paula Maria De Miguel, ESI Group, SPAIN Mr. Morgan Cameron, ESI Group, FRANCE Prof. Elias Cueto, Universidad de Zaragoza, SPAIN Prof. Francisco Chinesta, ENSAM, FRANCE Mr. Jean-Louis Duval, ESI Group, FRANCE