Localization method of hydrogen leak source for hydrogen supply cabin of fuel cell truck based on emendatory trilateration and model inversion

Fuel cell trucks frequently experience hydrogen system leaks due to vibrations and collisions of vehicles, making manual inspections for leak source localization time-consuming. Hydrogen system leakage within the hydrogen supply compartment of fuel cell trucks is focused in this paper. A physical model of a hydrogen storage compartment was constructed, measuring 950 mm in length, 1800 mm in width, and 1700 mm in height, along with an associated hydrogen supply system. Eight different environmental conditions were simulated, and hydrogen concentration data at six different points within the space were obtained using hydrogen concentration sensors. Based on the collected data, the study focused on localizing the hydrogen leak source. Initially, the Emendatory Trilateration Algorithm (ETA) was employed using a gas turbulence model to determine the approximate area and initial position of the leak source. The results showed that the identified area had a 75% probability of containing the actual leak source, and the average absolute error in the initial localization was 67.55 mm. Subsequently, the study combined a Gaussian puff model, which accumulates over time, with an Adaptive Crossover Rate-Differential Evolution Algorithm (ACR-DEA). The initial population was selected from the area determined by the ETA. The results demonstrated that the adaptive crossover rate effectively increased the algorithm's iteration speed, yielding an average error in the reverse calculation of the source strength of 2.684 x 10-4 kg/s and an average absolute error in the final localization of 41.65 mm.