Hydrogen Level Detection via Thermal Conductivity Measurement Using Temporal Temperature Monitoring

Hydrogen concentration measurement is important in fuel cells. Since the concentration of hydrogen fuel in these systems exceeds 50 v%, the mechanisms proposed for this task are inadequate due to their limited dynamic range or sensitivity at high concentration levels. Hence, more effective methods are sought after for the determination of H
₂
concentration at levels higher then ~10 v%. Here, we report the fabrication of a selective hydrogen sensor capable of measuring H
₂
levels in a wide dynamic range by recording the temporal variation of temperature on a solid body in equilibrium with the hydrogen containing atmosphere. The method is based on the fact that hydrogen has the highest thermal conductivity among other gases. An experimental database on these transient behaviors at different atmospheres is compiled. Feature vectors are constructed by the Padé z-transformation of the recorded temporal variations, and a neural network is designed and trained to relate the extracted features to the concentration level of the hydrogen. The system can measure the presence of hydrogen in concentration ranges of 40 to 100 % with acceptable accuracy.