Design, modeling and simulation study of a cascaded optimal neural network based fuel cell Powered Electric Vehicle

In this paper, the performance analysis of the ANN (Artificial Neural Network) based fuel cell powered electric vehicle is investigated for the two popular drive cycles such as M-UDDS and M-NEDC. The complex mathematical model of the fuel cell system is substituted with the black box neural network model that provides an appropriate mapping between the input and output parameters. The performance comparison of the two different cascaded connected neural networks is carried out to examine the prediction ability of the proposed network models in terms of error minimization value and convergence rate. The optimum network acquired from the comparative analysis can be used as an ancillary model instead of using a complex fuel cell model for developing any kind of fuel cell powered application. An attempt is made in this paper to use the neural network based fuel cell approach in the transportation sector for developing an electric vehicle model. This paper is also focused on the design, modeling and simulation of the optimal ANN based fuel cell operated electric vehicle and the performance of the proposed electric vehicle model is analyzed based on the two different drive cycle (M-UDDS & M-NEDC) on which they are operated. The simulation results obtained from the proposed electric vehicle model are used to evaluate the vehicle performance in terms of maximum distance coverage, amount of fuel consumption and comparison of the required vehicle power with the available power delivered by the energy source for the use of modified UDDS & NEDC drive cycle pattern. The power comparison results thus obtained enables to validate the optimality of the neural network model proposed in this paper.