Data‐driven H∞ loop‐shaping controller design

Summary We develop a data‐driven and real‐time test for candidate controllers' ability to meet given McFarlane–Glover loop‐shape specifications. The proposed test provides a data‐driven method to improve robustness of closed‐loop systems by eliminating controllers that past collected data have proved to fail to meet loop‐shape specifications. The test has an important property that many candidate controllers (not necessarily active ones) can be tested in real time using evolving real‐time plant data collected while others are active. It is shown that controller‐shaping in the present work plays a role that is dual to the role of plant‐shaping in the model‐based H∞ loop‐shaping design procedure. A simulation example is provided. Copyright © 2016 John Wiley & Sons, Ltd.