Solving the inverse Stefan design problem using genetic algorithms

Abstract

The article presents an algorithm of a solution of the two-phase inverse Stefan design problem. Based on the given information about the partial position of a moving interface, a functional was built defining the error of an approximate solution. To find the functional's minimum, a genetic algorithm was used, whereas for solution of the Stefan problem, the alternating phase truncation method was applied. Stability of the procedure was guaranteed using the Tikhonov regularization method. The results obtained corroborate the usefulness of the presented approach.