Data-driven dissipative verification of LTI systems: multiple shots of data, QDF supply-rate and application to a planar manipulator

TL;DR

This paper introduces a data-driven method for verifying dissipativity in LTI systems using multiple datasets, validated on a planar manipulator, advancing beyond single-shot data approaches.

Contribution

It proposes a novel dissipative verification technique leveraging multiple input-output datasets and quadratic supply-rate functions, applicable to practical robotic systems.

Findings

Successfully verified dissipativity of a planar manipulator using multiple data sets.

Demonstrated the method's effectiveness over recent single-shot data approaches.

Validated the approach in a real-world robotic system.

Abstract

We present a data-driven dissipative verification method for LTI systems based on using multiple input-output data. We assume that the supply-rate functions have a quadratic difference form corresponding to the general dissipativity notion known in the behavioural framework. We validate our approach in a practical example using a two-degree-of-freedom planar manipulator from Quanser, with which we demonstrate the applicability of multiple datasets over one-shot of data recently proposed in the literature.