Identification for stable second-order nonlinear systems

doi:10.1016/j.sysconle.2026.106427

Identification for stable second-order nonlinear systems

2026-07-01

0

PRE

AI

DOI：10.1016/j.sysconle.2026.106427

Original

Original request for help

Save

Abstract

En 中文

System identification is one of the fundamental techniques in the control society. However, although various identification approaches have been developed, we find that very rare works focus on naturally stable nonlinear systems, especially when systems are second-order. These systems are common in practice, such as closed-loop robot position and velocity tracking systems. Identifying these systems can benefit higher-level policies design, such as planning in the kinematic level. In this work, we present a learning-based approach to identify second-order nonlinear systems with a guarantee of global asymptotic stability. Inspired by the wellknown inertial-damping-spring system, we develop a general and flexible parameterized form for second-order nonlinear systems under a Lyapunov function learning framework. We show that if the derived learning constraints are satisfied, the property of global asymptotic stability for the identified system can always be ensured.

Keywords:

System identification

Second-order system

Globally asymptotic stability

Lyapunov function learning

Identification for stable second-order nonlinear systems

Abstract

Journal

Organization