Neural Network Modeling of Direct Ammonia Alkaline Anion Exchange Membrane Fuel Cell

Recently, there has been a growing interest in green ammonia as a feasible hydrogen-based alternative for clean energy generation. Ammonia presents various advantages such as simplified transportation and storage, higher density, and reduced flammability. Since ammonia can be used to directly generate electricity through a Direct Ammonia alkaline anion exchange membrane Fuel Cell (DAFC), an adequate model of DAFC is essential for operational studies and controller design. This paper proposes a novel neural network (NN) model for DAFC. The proposed NN model utilizes the input-output relationships of the DAFC to capture system dynamics, resulting in a rapid and highly accurate model. Traditional physics-based modeling of complex systems involving electrochemical reactions is arduous, time-consuming, and computationally intensive. Once trained and validated, this NN model is utilized to analyze the dynamic behavior of the DAFC. The research demonstrates that the NN-based approach is adept at predicting output parameters. The ability to develop precise data-driven models using solely data from validated physics-based models is demonstrated by the presented DAFC NN model, thus minimizing the need for extensive experimentation.