Characteristics of servo system using high order disturbance observer

Abstract

It is desirable to realize a nominal system which can control acceleration in order to realize fast and precise servo system, even if servo system has parameter variation and suffers from disturbance. We have shown that compensating disturbance and parameter variation by using disturbance observer can realize an ideal linear system which can control acceleration. In this paper, first, we transform a plant with disturbance compensation loop by disturbance observer into a two-degrees-of-feedom control system with a feedforward controller, which indicates that the disturbance observer system includes a series compensator such as P, PI, etc. depending on order of the disturbance observer. By clarifying the feature of the series compensator, we describe the bandwidth and the stability (the attenuation) of the system and show some simulation results, considering influence of the measurement noise.

Reduced order type-k disturbance observer based on a generalized Q-filter design scheme

“…Now, we call a DOB with (9) as 'reduced order type-k DOB'. Remark 2: In order to reject the polynomial-in time disturbance, a type-k DOB has been already pro posed in [8,9,12]. However, even if deg(Nn) = 0, the order of the type-k DOB with two Q-filters having iden tical structures is at least 2(v + k) although that of the proposed DOB with (9) is v + k + 1 .…”

“…But, it is conservative since it is derived by the small-gain theorem. Moreover, as the degree of Q-filter's numerator grows (i. e. , k increases), this condition tends to be violated [8]. Whereas, by a design procedure which will be proposed shortly, one can always design the re duced order type-k DOB to guarantee robust stability even though uncertain parameters of the plant belong to an arbitrarily large (but bounded) set.…”

“…TS) 2 +0.0675( TS) +0.0035 QD, p ,ty pe-2 (S) = (TS ) 3 + 1.4( 78) 2 +0.0675( 78) +0.0035 ' 1 QN, p , type-2 (S) = (78) 2 + 2(TS) + l 'To compare with the proposed design procedure, we select two Q-filters with binomial coefficients, which have been usually employed in the conventional DOB structure[8], as follows: QN,b(S) = QN, p , type-2 (S) . Notice that, in this case, pjd(S) is computed as Pjd,b(S) = (s + 1) 2 {s 3 +…”

High‐order variable‐structure disturbance estimators with applications to harmonic drive systems

“…Applications of the high-order linear disturbance estimators to servomechanisms have also been reported in Lee and Tomizuka (1996) and Kempf and Kobayashi (1999). It is shown in Yamada et al (1996) and Komada et al (2000) that a system can have low sensitivity when the order of the disturbance estimator is high enough. Otherwise, the disturbance affects the system when the order of the disturbance estimator is low (Komada et al, 2000).…”

Design of an Infinite-Order Disturbance Observer Enhancing Disturbance Suppression Performance

“…According to the Q-filter, high-order DOBs have been studied by considering specific frequencies (11) . A periodic disturbance is usually compensated by high-order DOBs by ignoring high-frequency harmonics (12) (13) . In contrast to the high-order DOBs, an infiniteorder disturbance observer (IFDOB) has been studied to suppress a periodic disturbance by considering all frequencies of the periodic disturbance (14) .…”