A Regularization and Active Learning Method for Identification of Quasi Linear Parameter Varying Systems

TL;DR

A manifold-regularization strategy is introduced that enforces smooth variations in the qLPV dynamics, promoting Linear Time-Varying (LTV) behavior and improving the extrapolation properties of the identified model.

Abstract

This paper proposes an active learning method for designing experiments to identify quasi-Linear Parameter-Varying (qLPV) models. Since informative experiments are costly, input signals must be selected to maximize information content based on the currently available model. To improve the extrapolation properties of the identified model, we introduce a manifold-regularization strategy that enforces smooth variations in the qLPV dynamics, promoting Linear Time-Varying (LTV) behavior. Using this regularized structure, we propose a new active learning criterion based on path integrals of an inverse-distance variance measure and derive an efficient approximation exploiting the LTV smoothness. Numerical examples show that the proposed regularization enhances qLPV extrapolation and that the resulting active learning scheme accelerates the identification process.