Heuristic Optimization Methods for the Characterization of Dynamic Mechanical Properties of Composite Materials

Abstract

Generally speaking, of the fundamental dynamic mechanical properties - mass, damping, and stiffness, damping is usually the most difficult to quantify. This is perhaps particularly true for composite materials which tend to have substantially higher damping than comparable isotropic materials and therefore having an accurate representation is correspondingly more important. Accordingly, some heuristic optimization techniques for the identification of the dynamic characteristics of honeycomb-core sandwich composite materials have been suggested. More specifically, Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) have been used in the present work as the optimization techniques for the system identification process. Experimental measurements of the dynamic responses (in the form of hysteresis loops) of simply-supported composite beam samples have been carried out, and a simplified semi-empirical mathematical model has been developed for such a system, tailored from individual experimental observations of the dynamic behavior of the samples when they are excited at their mid-points by sinusoidal displacement waves. The hysteresis loops that were obtained are for several frequencies and excitation amplitudes around the first mode of vibration. The analytical model contains four unknown system parameters, which must be identified by both optimization techniques utilized. The performance of these optimization methods are compared with computer-generated and experimental hysteresis loops. In addition, the effect of noise contamination in the signals has been studied in order to assess the search accuracy of the optimization algorithms under such conditions.