Robust Model of PEMFC for Mobypost Electric Vehicles
Article Sidebar
Published
May 16, 2016
Download
Statistic
Main Article Content
Abstract
Fuel Cell Hybrid Electric Vehicles (FCEV) are low emissions. However, these systems are still not very suitable for onboard applications because of their slow dynamic response and complex architecture. Nevertheless, an analytical model can be a suitable solution for improving the utilization of the fuel cells onboard in the electric vehicles. This model can replace the actual system, and thus reduce the instrumentation cost.
In the present paper, a signal based PEMFC model is proposed which is used to power a post-delivery hybrid electrical vehicle. This model is based on evaluating the variation effect of four parameters: temperature, mass flow, pressure and voltage. The objective of this modelling is to propose a simple identification technique, namely the AutoRegressive model with eXagenous inputs (ARX) and the Wavelet Transform (WT) in order to generalize a voltage variation model for a specific application such as FCHEV post-delivery vehicles. This model is simulated by using actual data. The simulation results are then compared with the experimental data from different vehicles. The results were found to be very promising (less than 2.5% of error) and show the robust model.
In the present paper, a signal based PEMFC model is proposed which is used to power a post-delivery hybrid electrical vehicle. This model is based on evaluating the variation effect of four parameters: temperature, mass flow, pressure and voltage. The objective of this modelling is to propose a simple identification technique, namely the AutoRegressive model with eXagenous inputs (ARX) and the Wavelet Transform (WT) in order to generalize a voltage variation model for a specific application such as FCHEV post-delivery vehicles. This model is simulated by using actual data. The simulation results are then compared with the experimental data from different vehicles. The results were found to be very promising (less than 2.5% of error) and show the robust model.