SearcharxivSearch

arXiv subjects

Herman

Publications and source records attributed to Herman.

2 recordsLinked to original sources

Combustion Behaviour of Single Silicon Particles in Different Oxidizing Environments

Silicon, despite its abundance and high energy density, remains underexplored as a carbon-free fuel, with limited data available on its combustion characteristics. In this work, the combustion behaviour of silicon particles is examined using an electrostatic levitator with laser ignition. Five oxidizing environments at atmospheric pressure are investigated: air, pure oxygen, and mixtures containing 40% oxygen (by mole) diluted with nitrogen, helium, or argon. The burning droplet peak temperature, measured by three-colour pyrometry, increases by 337 K from air to pure oxygen. The peak temperature of the silicon droplet in the 40%O2-60%He mixture is lower than that in the 40%O2-60%Ar mixture, in contradiction with thermodynamic predictions, due to a higher Lewis number of the helium-diluted mixture. Although oxygen diffusivity is higher in the helium-diluted mixture, a lower burning rate is observed, attributed to the lower combustion temperature. High-speed colour camera observations reveal that the square of the particle diameter decreases with time in each combustion run, following a strong linear relationship (R2 > 0.99) across all oxidizing environments. However, the combustion lifetime is proportional to the initial particle diameter to the power of n, with n ranging from 1.69 to 1.82. This deviation from the expected n = 2 appears to result from unavoidable measurement uncertainties and the limited particle size range, rather than differences in combustion physics. The decrease in silicon droplet size during combustion is attributed to the formation of gaseous SiO as an intermediate combustion product. The SiO species is observed using a UV camera, showing a UV intensity decay from the particle surface. High-speed imaging and LED absorption signals indicate that the final condensed product, SiO2 nanoparticles, are not optically visible, suggesting they possess very low emissivity.

physics.app-ph

Transfer-Learning-Aware Neuro-Evolution for Diseases Detection in Chest X-Ray Images

The neural network needs excessive costs of time because of the complexity of architecture when trained on images. Transfer learning and fine-tuning can help improve time and cost efficiency when training a neural network. Yet, Transfer learning and fine-tuning needs a lot of experiment to try with. Therefore, a method to find the best architecture for transfer learning and fine-tuning is needed. To overcome this problem, neuro-evolution using a genetic algorithm can be used to find the best architecture for transfer learning. To check the performance of this study, dataset ChestX-Ray 14 and DenseNet-121 as a base neural network model are used. This study used the AUC score, differences in execution time for training, and McNemar's test to the significance test. In terms of result, this study got a 5% difference in the AUC score, 3 % faster in terms of execution time, and significance in most of the disease detection. Finally, this study gives a concrete summary of how neuro-evolution transfer learning can help in terms of transfer learning and fine-tuning.

cs.NE