Identification for gain-scheduling: a balanced subspace approach

Abstract

Abstract-The problem of deriving MIMO parameterdependent models for gain-scheduling control design from data generated by local identification experiments is considered and a numerically sound approach is proposed, based on subspace identification ideas combined with the use of suitable properties of balanced state space realisations. Simulation examples are used to demonstrate the performance of the proposed approach.

Combining Analytic and Experimental Information for Linear Parameter-Varying Model Identification: Application to a Flexible Robotic Manipulator

“…Notice that this dynamic evolution totally differs from the one satisfied by the local black-box balanced matrices A B C Fig. 6) which was the case in [22] as well. This comparison shows that, at least locally, combining structural information with black-box estimates can lead to performance equivalent to the one reachable through a global analytic study which requires strong knowledge in robotics as well as the availability of all of the physical parameters governing the behavior of the system.…”

Linear parameter varying modeling and identification for real-time control of open-flow irrigation canals

“…(Lee, 1999;Bamieh, 2002;Verdult, 2005;Felici, 2007) and a local (see e.g. (Steinbuch, 2003;Wassink, 2005;Lovera, 2007;Paijmans, 2008;De Caigny, 2009) one.…”

Grey-box modeling of an inverted pendulum system via identification of a PD-LTI system

“…In the local approach, local LTI (Linear Time-Invariant) models are identified for different fixed operating points. For choosing the adequate state in interpolating state-space models, balanced realization has been used by Lovera and Mercère [7], and the SMILE technique has been proposed by De Caigny et al [8]. Grey-box modeling approach has been moreover taken by Tanaka et al [9,10], and a PD-LTI (Parameter-Dependent Linear Time-Invariant) system has been introduced by Tan et al [10].…”

“…[6]) and local (e.g. [7,8]) approaches. In the local approach, local LTI (Linear Time-Invariant) models are identified for different fixed operating points.…”