Flow Structure Identification for Nonlinear Dynamical Systems via Finite-Time Lyapunov Analysis

Department of Mechanical and Aerospace Engineering, University of California Irvine CA, 92697, USA

^⋆ Corresponding author. E-mail: kmease@uci.edu

Abstract

Identifying and characterizing geometric structure in the flow of a nonlinear dynamical system can facilitate understanding, model simplification, and solution approximation. The approach addressed in this paper uses information from finite-time Lyapunov exponents and vectors associated with the tangent linear dynamics. We refer to this approach as finite-time Lyapunov analysis (FTLA). FTLA identifies the potential for flow structure based on the stability and timescales implied by the spectrum of finite-time Lyapunov exponents. The corresponding Lyapunov vectors provide a basis for representing a splitting of the tangent space at phase points consistent with the splitting of the spectrum. A key property that makes FTLA viable is the exponential convergence of the splitting as the finite time increases. Tangency conditions for the vector field are used to determine points on manifolds of interest. The benefits of the FTLA approach are the dynamical model need not be in a special normal form, the manifolds of interest need not be attracting nor of known dimension, and the manifolds need not be associated with a fixed point or periodic orbit.

After a brief review of FTLA, it is applied to spacecraft stationkeeping around a libration point in the circular restricted three-body problem. This application requires locating the stable and unstable subspaces at points on periodic and aperiodic orbits. For the periodic case, the FTLA subspaces are shown to agree with the Floquet subspaces; for the quasi-periodic case, the accuracy of the FTLA subspaces is demonstrated by simulation.