Improving the efficiency of extended state observer under noisy measurements by low-pass filter

Abstract

In this paper, with an attempt to extract a purified signal from the output signal with measurement noise, a measurement noise attenuation technique is applied into the extended state observer (ESO). A first order low-pass filter is utilized to filter the high-frequency measurement noise. By doing so, ESO is able to estimate the total uncertainties more accurate, and avoids the danger of making the system be unstable due to noise amplification. Comparisons between ESO and the low-pass filter based ESO (LFESO) have been made via some typical simulation examples. Numerical results confirm that the considered technique is capable of improving the ability of noise attenuation greatly.