Performance of artificial neural networks in an inverse problem of laser beam diagnostics

Abstract

Results are presented for the numerical verification of a method devised to identify an unknown spatio-temporal distribution of heat flux that occurs at the surface of thin aluminum plate, as a result of pulsed, high-power laser beam excitation. The presented identification of boundary heat flux function is a part of newly-proposed laser beam profiling method and utilizes artificial neural networks trained on temperature distributions generated with the ANSYS Fluent solver. The paper focuses on the selection of the most effective neural network hyperparameters (Keras, Tensorflow) and compares the results of neural network identification with Levenberg-Marquardt method used earlier and discussed in our previous articles.

arXiv comment20 pages, 12 figures, 1 table, 61 references

References in corpus (5)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• An overview of gradient descent optimization algorithms
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Activation Functions: Comparison of trends in Practice and Research for Deep Learning
• Inverse analysis for the identification of temporal and spatial characteristics of a short super-Gaussian laser pulse interacting with a solid plate