Parameter Identification in Mechanical Systems with Energy-Based Regressor: Preliminary Study

Abstract

Parameter estimation in physical systems is an important area of modern engineering and robotics. This paper discusses parameter identification in linear, time-invariant mechanical systems based on their energy. The described approach employs system’s regressor derived from the energy equation with subsequent least-squares estimation. This method has been evaluated experimentally on a single degree-of-freedom pendulum and demonstrated high accuracy and quick parameter convergence, while not suffering from the drawbacks of conventional dynamics-based identification.