Comparative study of evolutionary algorithms for parameter identification of an impact oscillator

The use of non-classical evolutionary optimization techniques such as genetic algorithms, differential evolution, swarm optimization and genetic programming to solve the inverse problem of parameter identification of dynamical systems leading to chaotic states has been gaining popularity in recent years. In this paper, three popular evolutionary algorithms-differential evolution, particle swarm optimization and the firefly algorithm are used for parameter identification of a clearance-coupled-impact oscillator system. The behavior of impacting systems is highly nonlinear exhibiting a myriad of harmonic, low order and high order sub-harmonic resonances, as well as chaotic vibrations. The time-history simulations of the single-degree-of-freedom impact oscillator were obtained by the Neumark-b numerical integration algorithm. The results are illustrated by bifurcation graphs, state space portraits and Poincare' maps which gives valuable insights on the dynamics of the impact system. The parameter identification problem relates to finding one set of system parameters given a chaotic or periodic system response as a set of Poincaré points and a different but known set of system parameters. The three evolutionary algorithms are compared over a set of parameter identification problems. The algorithms are compared based on solution quality to evaluate the efficacy of using one algorithm over another.