Robustness improvement of existing model reference adaptive control in the presence of unknown time-varying dynamics

This paper designs a new robust model reference adaptive control (MRAC) architecture in the requirement of large adaptation gain to achieve fast adaptation for performance recovery. lt is well known that traditional standard MRAC with high-gain learning rate will result in high-frequency oscillations in the control signal and even system instability. To tackle this problem, we add a new term in the MRAC adaptive law and robustness is improvement. This modification term involves a low-pass filter to filter out high-frequency oscillations.The performance bound for both zero and nonzero initial errors are obtained through Lyapunov method. simulation results show this architecture guarantees fast adaptation without losing robustness. Furthermore, it has potential capability to compensate for input time-delays.