Abstract
Estimate of the l2 gain of systems by using a transient response is essential for unfalsified control. We have proposed an l2 gain estimation method using a filter-bank for efficient falsification for linear time-invariant systems, and developed a data-driven loop shaping method for off-line PID tuning. Estimate of the l2 gain is also essential for model validation, where the extension theorem gives the theoretical basis. In this paper, we clarify the property of our filter bank method comparing with the extension method, and the l2 gain estimates by these two methods are compared in accuracy, efficiency, and robustness against noise by numerical experiments.
