Modeling and performance analysis of neural network based fuel cell driven electric traction system

The paper is focused on design, modeling and simulation of the Neural Network (NN) based Proton Exchange Membrane (PEM) fuel cell driven electric vehicle system. The Multi Layer Perceptron (MLP) NN model is used to predict the output voltage of the PEM fuel cell for a predefined vehicle driving cycle pattern. The optimality of neural network is investigated with the four different training algorithms along with the varying hidden neurons. The performance comparison of the neural network is carried out to select the optimum network model in terms of error minimization values and convergence rate. The optimum network which is chosen is used to develop a neural network based PEM fuel cell driven electric vehicle that incorporates the modeling of neural network based fuel cell, DC-DC converter and vehicle dynamics. The modeling and simulation work is carried out using MATLAB/Simulink in order to verify the reliability of the electric vehicle performance by applying three different modified standard driving cycle (M-NEDC/M-UDDS/M-US06) patterns. The simulation results obtained from the developed electric vehicle model are used to evaluate the vehicle performance in terms of amount of fuel consumption, maximum distance coverage and power flow within the vehicle system.