Application of Artificial Neural Networks to Heat Transfer Simulations of Thin Film Structures Irradieted by Laser

In the presented work, heat transfer in one-dimensional numerical model of thin films irradiated by ultrashort laser pulses is discussed. In proposed mathematical model some laser parameters are given as interval numbers, which simulate the noise that can happen in reality. The direct problem has been solved using the rules of directed interval arithmetic. As numerical solution of such problem may be computationally demanding, the authors propose using artificial neural networks as fast approximators of interval results of important parameters for technological processes. In the numerical example, artificial neural network was trained from the numerical model to estimate the maximum temperature (as an interval number) and time when it is achieved. The accuracy and computational time of the problem solutioning using artificial neural network were measured and compared to lattice Boltzmann method.