Data-Based Model of PEM Fuel Cell Using Neural Network

A proton exchange membrane (PEM) fuel cell is an alternative energy source. Generally, the fuel cell is an electrochemical device in which the chemical energy is converted into electric energy, and the by-product of the fuel cell is water. Apart from the electrochemical reaction, multiple physical processes, such as heat and mass transfer, water formation, and vaporizations, occur during the function of the fuel cell. In the literature, various models based on multi-physics along with electrochemical reactions were developed to predict the performance of the fuel cell. In recent years, the machine learning and data-driven approach has been implemented in many applications to improve the study of the system. This paper presents the data-based model for the PEM fuel cells using an Artificial Neural Network (ANN) to anticipate the cell voltage for various operating conditions. The prediction of the fuel cell is done by collecting the data from the fuel cell using a data acquisition system. The backpropagation ANN technique is employed to predict the performance of the fuel cell by considering the three input parameters: temperatures, current density, and hydrogen gas pressure. The predicted response based on the ANN technique is obtained with a regression value > 0.99 for all the variables. A comparison of the actual output of the fuel cell and data-based model shows that the data-based model is more accurate and also reduces the need for wide experimentation in the physics-based models.