The Automatic Dynamic Penalisation method (ADP) for handling constraints with genetic algorithms

In this paper we present a new penalty-based approach, developed within the framework of genetic algorithms (GAs) for constrained optimisation problems. The proposed technique, which is called Automatic Dynamic Penalisation (ADP) method, belongs to the category of exterior penalty-based strategies. The aim of this work consists in providing a simple and effective constraint-handling technique without the need of tuning the penalty coefficients values for any considered optimisation problem. The key-concept that underlies the ADP strategy is that it is possible to exploit the information restrained in the population, at the current generation, in order to guide the search through the whole definition domain and to give a proper evaluation of the penalty coefficients. The proposed strategy is firstly applied to three different benchmark problems and the obtained results are compared to those available in the literature in order to show the effectiveness of the ADP technique. Finally, as examples of real-world engineering applications, the ADP method is employed to search a solution for two different optimisation problems, i.e. the optimal design of damping properties of hybrid elastomer/composite laminates and the maximisation of the first buckling load of composite laminates with given elastic symmetries.