Application of Artificial Neural Network Technologies in Control Systems of Power Plants Based on Open-Cathode Hydrogen Fuel Cells

This article presents research results examining the feasibility of using artificial neural network (ANN) technologies to predict the performance of power plants with open-cathode hydrogen fuel cells (FCs) under changing environmental conditions. Results of experiments conducted in a climate chamber to study the effect of ambient temperature on FCs with a polymer proton-exchange membrane are presented and used to develop databases for training the ANN. A recurrent ANN architecture for predicting FC voltage and hydrogen consumption is proposed. Training results confirm the feasibility of using ANN technologies in developing adaptive control systems for FC-based power plants.