Abstract— In this paper, a nonsmooth Kalman filtering (NKF)
strategy is proposed for state estimation of macro-motion
positioning stages with ball-screw-nut transmission mechanism.
In this method, the characteristic of the positioning stage is
described by so-called sandwich state-space model with
backlash-like hysteresis. By considering the influence of random
noise on positioning stage, based on the dynamic model with
backlash-like hysteresis, a nonsmooth Kalman filter is proposed
for state estimation of positioning stage in noisy environment.
Moreover, a method of observability test for sandwich systems
with backlash-like hysteresis is developed for observability
check before the nonsmooth Kalman filter is implemented.
Finally, the proposed NKF method is applied to a positioning
stage for state estimation.
I. INTRODUCTION
Macro-motion positioning stages are usually used in
machine tools and computer numerical control (CNC)
machine centers etc..[1]. In a macro-motion positioning stage,
a servo-motor is used to drive a work platform through a
transmission mechanism using gearbox or ball-screw-nut. To
describe the characteristic of the stage, we can use a linear
dynamic submodel to describe behavior of servo-motor,
which is called as the input linear dynamic submodel, and
employ another linear dynamic submodel to depict the
characteristic of work platform, which is regarded as the
output linear dynamic submodel. Meanwhile, in order to
describe the phenomenon of backlash-like hysteresis caused
by the gaps between mated gear teeth or gaps between
ball-screw and nut in transmission mechanism, a
backlash-like hysteresis submodel is developed, which is
embedded in between the input linear submodel and output
linear submodel. Thus, the macro-motion positioning stage
can be depicted by a sandwich model with backlash-like
hysteresis [2].[3]. Ref. [3] has proposed a recursive
identification strategy for sandwich systems with backlash.
This method has also been applied to identification of a
macro-motion positioning stage [3].
This work is partially supported by the international cooperation research
project of the Science and Technology Commission of Shanghai (Grant Nos.
14140711200, 14ZR1430300), and National Science Foundation of China
(NSFC Grant Nos.: 61203108, 61371145, 61171088 and 61571302).
Y. Tan, R. Dong, and H. He are with the College of Information,
Mechanical and Electrical Engineering, Shanghai Normal University,
Shanghai 200234, China (e-mail: tany@shnu.edu.cn;
dongrlnpu@shnu.edu.cn
, and heh@shnu.edu.cn).
Y. Li is with the Institute of Robotics and Automatic Information System,
Tianjin Key Laboratory of Intelligent Robotics, Nankai University, Tianjin
300071, China (e-mail:lyy0825@mail.nankai.edu.cn).
X. Chen is with the Department of Electrical and Computer Engineering,
University of Windsor, Windsor, ON, Canada, N9B 3P4
(xchen@uwindsor.ca).
On the other hand, the output of displacement sensor of the
stage may be contaminated by measurement noise e.g.
thermal noise in amplifier circuit, electromagnetic
interference and power-line noise etc.. Meanwhile, the
positioning stage is also disturbed by vibration of stage base,
gear meshing noise and bearing noise etc.. For the optimal
control as well as fault diagnosis of macro-motion positioning
stage disturbed by stochastic noise, development of an on-line
dynamic filtering strategy is necessary for noise elimination
and state estimation.
For linear systems in stochastic cases, Kalman filter (KF)
has been very useful for stochastic control systems [8].
However, the mentioned KF method has never related to the
dynamic sandwich systems with backlash-like hysteresis. The
extended Kalman filter can be used for noise suppression for
smooth nonlinear systems but is not applicable directly to the
macro-motion positioning stage with non-smooth
backlash-like hysteresis.
For state-estimation of positioning stages with
backlash-like hysteresis, Ref. [4] proposed a scheme to
construct non-smooth observer for sandwich systems with
backlash whilst Ref.[5] employed a non-smooth observer for
fault diagnosis of sandwich systems with backlash. However,
in those aforementioned strategies, the influence of random
noise is not considered.
In this paper, a non-smooth Kalman filter (NKF) is
proposed for state estimation of macro-motion positioning
stages with the property of backlash-like hysteresis. In order
to construct a non-smooth Kalman filter, a non-smooth
state-space sandwich model with backlash-like hysteresis is
introduced to describe the characteristic of macro-motion
positioning stage with backlash-like hysteresis. Afterwards, a
non-smooth Kalman Filter is developed based on the obtained
non-smooth state-space sandwich model.
It is known that a necessary condition for the Kalman Filter
to work correctly is that the system to be estimated is
observable. Therefore, in this paper, a method of
observability test for sandwich systems with backlash-like
hysteresis is proposed before the NKF is implemented
.
Then, the proposed NKF method is applied to the state
estimation of macro-positioning stage disturbed by stochastic
noise.
II. S
TATE-SPACE MODEL FOR POSITIONING STAGES
In terms of the architecture of a typical macro-motion
positioning stage, the DC servo-motor and work platform are
usually regarded as linear dynamic subsystems. Hence, we
use the input linear dynamic submodel and output linear
A Nonsmooth Kalman Filter for State Estimation of Positioning Stage
Based on Sandwich Model with Backlash-like Hysteresis
Yonghong Tan, Yanyan Li, Ruili Dong, Xiang Chen, Hong He
2016 American Control Conference (ACC)
Boston Marriott Copley Place
July 6-8, 2016. Boston, MA, USA
978-1-4673-8682-1/$31.00 ©2016 AACC 661