Modeling and Control of MCFC System Based on Artificial Neural Networks

Artificial neural network model of molten carbonate fuel cells: Validation on experimental data

“…4,6,8,10,15 There are also molten carbonate fuel cell models based on other assumptions, eg, reduced order model, 12,13 kinetic model, 14 and model using ANNs. [1][2][3] This paper presents a number of mathematical models based on ANNs to predict MCFC performance, though it also draws on knowledge gained from the use of ANNs in modeling a whole range of other devices.…”

“…Among the listed construction parameters, we have only adequate experimental data (1)(2)(3). The influence of the electrolyte material can be tested using a hybrid model, based in part on known dependencies for the processes that occur here (electrolyte conductivity as a function of temperature for different materials).…”

“…Most MCFC models are based on the Nernst equation and include losses of activation, concentration, and ohmic losses or losses on the anode, electrolyte, and cathode . There are also molten carbonate fuel cell models based on other assumptions, eg, reduced order model, kinetic model, and model using ANNs …”

“…Tian and Zou presents an MCFC fuel cell model based on ANN, and a control system based on this model was built. The type of ANN used was the internal recurrent neural network (IRN), with the structure 7‐5‐2.…”

“…Several papers report on the use of ANNs for various purposes pertaining to fuel cell technology (see Section ). Three in particular deal with molten carbonate fuel cells (MCFCs) while others are based on experimental data taken from the literature. In this paper, we use ANNs to simulate the behavior of MCFCs, basing the training process on our own experimental data, which makes the work unique.…”