doiSerbia

Experimental verification of approximate solution of the inverse Stefan problem obtained by applying the invasive weed optimization algorithm

Hetmaniok Edyta (Silesian University of Technology, Institute of Mathematics, Gliwice, Poland)
Slota Damian (Silesian University of Technology, Institute of Mathematics, Gliwice, Poland)
Zielonka Adam (Silesian University of Technology, Institute of Mathematics, Gliwice, Poland)

The paper proposes a procedure for solving the inverse Stefan problem consisted in reconstruction of the function describing the heat transfer coefficient on the basis of temperature measurements. Elaborated method is based on two procedures: solution of the appropriate direct Stefan problem by using the finite difference method combined with the alternating phase truncation method and minimization of some functional with the aid of invasive weed optimization algorithm. For verifying the effectiveness of investigated algorithm the experimental data obtained in the solidification of aluminum are used.

Keywords: artificial intelligence, evolutionary algorithms, invasive weed optimization algorithm, inverse Stefan problem