An Approach to Sparse Continuous-time System Identification from Unevenly Sampled Data

TL;DR

This paper introduces a novel method for identifying sparse continuous-time dynamical systems from unevenly sampled data, combining cross-validation and subset growth to select relevant basis functions.

Contribution

It presents a new approach that effectively handles irregular sampling in system identification by integrating machine learning techniques with iterative subset selection.

Findings

Successfully applied to a 6-node feedback ring

Effective on Van der Pol oscillator data

Outperforms traditional methods in irregular sampling scenarios

Abstract

In this work, we address the problem of identifying sparse continuous-time dynamical systems when the spacing between successive samples (the sampling period) is not constant over time. The proposed approach combines the leave-one-sample-out cross-validation error trick from machine learning with an iterative subset growth method to select the subset of basis functions that governs the dynamics of the system. The least-squares solution using only the selected subset of basis functions is then used. The approach is illustrated on two examples: a 6-node feedback ring and the Van der Pol oscillator.