Hydrogen leakage identification of hydrogen fuel cell vehicles in underground garages using Tikhonov regularization and Bayesian methods

The improvement in hydrogen fuel cell vehicles markedly aids carbon reduction in transportation. Nevertheless, hydrogen is a highly combustible and explosive gas. Addressing the potential risk that hydrogen leakage from these vehicles in underground garages may pose to the garage structure and occupants, the reliability of the hydrogen leakage model is validated firstly through experimental data. Subsequently, the simulation results of the hydrogen leakage model are integrated with Tikhonov regularization and Bayesian methods to develop an identification model. The model can identify the release rate profiles with constant and decaying release. Finally, the impact of sensor distribution positioning on both the release rate and location identification is determined. It is shown that the relative error in identifying the constant release rate linked to the optimal regularization parameter ranges from 0.32% to 5.80%. Accurately locating a leakage depends more on the average relative error than on sensor arrangement. When the average relative error in predicted concentration is less than 0.03 %, sensors positioned at various locations can correctly identify the leak's location with a median probability of approximately 75%. This research can provide a reliable source information about the occurrence of unexpected hydrogen leaks.