Evolutionary algorithms for propellants parameters identification

This paper presents the possibility of the parametric identification of an unknown low energy explosive basing only on its ballistic curve using advanced optimisation algorithm. What was under investigation here was the approach of an explosive replication for the numerical investigation of ANSYS AUTODYN solver. The emphasis was given here for the propellants used in the automotive safety devices. The results of closed bomb testing was compared with its numerical representation. The required parameters to develop a new numerical model of an explosive was identified with the aid of evolutionary algorithm. In order to perform a parametric identification a mathematical model of considered phenomenon is required. Hence, this paper contains a mathematical model of a deflagration process which was the basis for the evolutionary algorithm. The algorithm verified a variety of parameters until the objective function is obtained. In the case of this paper the objective function was a ballistic curve of an unknown explosive which combust under deflagration regime. The results obtained with this method shows good agreement with the closed bomb test of the propellant. Furthermore, advanced optimisation tools such as an evolutionary algorithms, in oppose to most of other optimisations algorithms, enables to find a global optimum. However, the identified function here was found to be unimodal.