H∞-based LPV model identification from local experiments with a gap metric-based operating point selection

When the identification of linear parameter-varying (LPV) models from local experiments is considered, the question of the necessary number of local operating points usually arises. In this work, this challenging problem is tackled by proposing a method which is able to optimize on-line the number of local operating points required by the local technique used to identify the LPV model parameters. This goal is achieved by developing an algorithm which takes the advantage of the gap metric-based non-linearity measure [1]. The proposed method is then embedded to an H_∞-based technique and tested by identifying a fully-parameterized and a physically-structured LPV model written as a linear fractional representation (LFR).