抄録
In the design of control systems and measuring instruments, an input estimator is required if the disturbance input is inaccessible.
This input estimator becomes an inverse system as is well-known if the disturbance input is completely unknown. This is derived from the theory of the unknown input observer which estimates the whole state variables of plants with completely unknown inputs. However this inverse system requires several differentiators and its estimating speed can not be set arbitrarily since the poles become the zeros of the plant.
Assuming that the disturbance input is given by the polynomial of order p with respect to time, Meditich et al. have introduced the p-observer which estimates the state variables of the plant. This paper derives conditions under which the p-observer estimates only the disturbance input and shows the following results.
1) The p-observer requires no differentiators and its poles can be assigned arbitrarily. Using the transfer function from the disturbance input to its estimate, the role of this p-observer is clarified in view of the classical control theory.
2) Assuming that the swift modes of the plant are included in the observer, a new input estimator called a quasi-observer is proposed such that its dynamical order is far less than the minimal one obtained from the observer theory.
