A Bi-bandwidth Extended State Observer for System with Measurement Noise

Although traditional Extended State Observer (ESO) can efficiently estimate the total disturbance of the system theoretically, observation may diverge from the actual system when the measurement noises exist, due to the sensibility of such high-gain-observer to noises. The trade-off between transient property and denoise capacity of ESO considerably complicates the tuning of gain parameters. Taking this into account, we proposed a novel ESO-based approach, the Bi-bandwidth ESO (BESO), which uses direction information of estimation error rate to tune the bandwidth of ESO dynamically. Numerical comparisons among the LESO, NESO, and BESO, are given to demonstrate the effectiveness of our proposed approach for tackling observation issues with measurement noises.