Configuring genetic algorithm to solve the inverse heat conduction problem

Solving the inverse heat conduction problem is the key to determining heat flux on the surface of an object. Unfortunately, it is a typical ill-posed problem, as only numerical approximation methods are available for finding an approximate solution. It is common to use some heuristic methods, such as genetic algorithms, particle swarm optimisation, etc. Although the main mechanisms of these approaches are well-known, the practical implementations raise several questions about the free parameters (population size, elitism, mutation probability/range, etc.). This paper presents the results of several experimental tests to find the appropriate attributes of a genetic algorithm based approach to quickly and reliably solve the inverse heat conduction problem.