Construction of the stationary probability density for a family of SDOF strongly non-linear stochastic second-order dynamical systems

Abstract

In this paper, a new procedure is proposed to construct the stationary probability density for a family of the single-degree-of-freedom (SDOF) strongly non-linear stochastic second-order dynamical systems subjected to parametric and/or external Gaussian white noises. First of all, the Fokker–Planck–Kolmogorov (FPK) equation associated with the original Itô stochastic differential equation is replaced by the equivalent FPK equation by adding arbitrary anti-symmetric diffusion coefficient. Then, a family of invariant measures depending on the arbitrary anti-symmetric diffusion coefficient and another arbitrary function is constructed by vanishing the probability flows in two directions. Finally, the drift vector associated with a family of Itô stochastic differential equations is deduced by giving, a priori, these two arbitrary functions. It is shown that the known invariant measures dependent on energy are only the special cases of invariant measures presented in this paper, while some other classes of invariant measures are independent of energy. Thus, the invariant measures constructed in this paper are those belonging to the most general class of the SDOF strongly non-linear stochastic second-order dynamical systems so far.