Global Analysis of Stochastic Systems by the Digraph Cell Mapping Method Based on Short-Time Gaussian Approximation
The digraph cell mapping method is popular in the global analysis of stochastic systems. Traditionally, the Monte Carlo simulation is used in finding the image cells of one-step mapping, and it is notably costly in the computation time. In this paper, a novel short-time Gaussian approximation (STGA) scheme is incorporated into the digraph cell mapping method to study the global analysis of nonlinear dynamical systems under Gaussian white noise excitations. In order to find out all the active image cells in one-step cell mapping quickly, the STGA scheme together with a probability truncation method is introduced for systems without periodic excitation, and then in the case with periodic excitation. The global structures, such as the stochastic attractors, stochastic basins of attraction and stochastic saddles, are calculated by the digraph analysis algorithm. The proposed methodology has been applied to three typical stochastic dynamical systems. For each system, the effectiveness and superiority of the proposed STGA scheme are verified by checking the image cells of one-step mapping and comparing with the results of Monte Carlo simulation. It is found in the global analysis that the change of the amplitude of periodic excitation induces stochastic bifurcations in the stochastic Duffing system. Moreover, a stochastic bifurcation occurs in the stochastic Lorenz system with the increase of noise intensities.