Peaking Reduction of CRM-Based Adaptive Control via a Modified Adaptive Law

Abstract

To circumvent the oscillations induced by high-gain adaptation in the model reference adaptive control (MRAC), closed-loop reference model (CRM) based MRAC was recently proposed. However, although a fair steady-state performance can be guaranteed, the induced peaking phenomenon in the CRM based adaptive control system may deteriorate the transient tracking response. In this paper, we first analyze the peaking value based on L2 norm and Cauchy-Schwartz inequality. Then according to the analysis, a novel adaptive law containing the parameter estimation error is constructed to alleviate the peaking phenomenon. This proposed method can allow using a fairly large feedback gain in the CRM based MRAC system to achieve better transient performance. Experiment results based on a 3-DOF helicopter show that the modified CRM adaptive control system can alleviate the peaking phenomenon.