Improved Cascade Equivalent-Input-Disturbance Approach to Rejecting Disturbance With Different Frequencies

The equivalent-input-disturbance (EID) approach has been presented and used extensively to actively estimate and compensate for unknown disturbances. To ensure satisfactory performance, the bandwidth of the EID estimator should be chosen to encompass the entire frequency range of a disturbance, which makes the estimator extremely sensitive to measurement noise when the disturbance contains a high-frequency component. This leads to a tradeoff between disturbance rejection and noise suppression. This study presents a new method to address this issue. It uses two EID estimators in the form of a cascade structure to handle a disturbance with a slowly varying component and a high-frequency component separately. The structure of an improved cascade EID-based motion-control system is described. A band-pass filter is employed in one estimator to handle the high-frequency component, while a low-pass filter is applied in another estimator to deal with the slowly varying component. A frequency-domain analysis reveals that the control system’s sensitivity for disturbance rejection is improved, while its complementary sensitivity for noise suppression remains nearly unchanged at high frequencies. It also shows that the design of the two estimators is independent of each other. The stability criteria for the system are derived. Experimental results show that the two components are estimated separately and effectively suppressed from the output of the system. The presented method provides a promising technique for simultaneously rejecting disturbances and suppressing measurement noise. It also has the potential to distinguish disturbances with different frequencies, which can be applied to fault diagnosis.