Efficient elimination of offset, in the pole-assignment self-tuning feed-back control system is discussed. Offset can be eliminated by integrator and it is shown that more efficient elimination of offset can be achieved by the cancellation of stable zeros with the poles of controller. This approach also improves various properties such as set-point following characteristics and reduced control input. Efficient offset elimination is achieved by the analysis of noise term in relation with the cancellation of stable zeros which is previously never consideres with severity.
Conversion of the proposed pole-assignment controller to the discrete PID form is also performed for the more popular application in practice. Also the ill-conditioning problem which arise due to the exact or near pole-zero cancellation in the plant model, is avoided by the cancellation of stable zeros.
A method to measure sensitivity to the modeling errors is suggested and proved based on the estimation of power spectrum and the usefulness of the method is checked in the example.