Stabilization of an Underactuated BABS Using a SOSM Control
Paper:
Stabilization of an Underactuated Ball-and-Beam System
Using a Second-Order Sliding Mode Control
Jie Yang
∗
, Qinglin Wang
∗
,YuanLi
∗
, and Jinhua She
∗∗
∗
School of Automation, Beijing Institute of Technology
5 South Zhongguancun Street, Haidian, Beijing 100081, China
E-mail: yangjieyangjie1985@126.com, wqlbit@gmail.com, liyuan@bit.edu.cn
∗∗
School of Computer Science, Tokyo University of Technology
1401-1 Katakura, Hachioji, Tokyo 192-0982, Japan
E-mail: she@cs.teu.ac.jp
[Recei ved May 22, 2013; accepted January 4, 2014]
This paper presents a stabilization method for an un-
deractuated ball-and-beam system (BABS) based on
a second-order sliding mode (SOSM) control. The
BABS is an underactuated nonlinear system that is
widely used to verify nonlinear control performance.
Virtual control is introduced to a second-order BABS
subsystem to m inimize control performance inaccu-
racy by using model linearization. An actual vir-
tual controller with variable finite-time tracking is
achieved using a second-order sliding mode controller.
An adaptive robust method is proposed to solve an un-
certainty problem with unknown upper bounds, and
then a finite-time convergence theory proof is given.
Theory, simulation and experiment results verify the
efficiency of the BABS controller.
Keywords: ball-and-beam, second-order sliding mode,
uncertainty, finite-time convergence
1. Introduction
Sliding mode control (SMC) is a robust control ap-
proach widely used for uncertain systems. Conventional
SMC requires that the relative degree of the sliding sur-
face be 1 and that the effect of control appear clearly in
the first derivative of the sliding surface. A significant
disadvantage of conventional SMC is the appearance of
chattering [1]. A higher-order sliding mode (HOSM) con-
troller is developed to reduce the effect of chattering on
system output while maintaining SMC robustness, remov-
ing relative degree 1 constriction, and improving control
accuracy. An HOSM controller does not work directly
on the derivative of a sliding surface but on the second-
or higher- order derivative surface. The most widely
used HOSM is the second-order sliding mode (SOSM).
Since the SOSM controller is easy to implement and pro-
vides better control accuracy than the conventional slid-
ing mode controller, it has attracted considerable attention
over the last decade [2].
An underactuated system has fewer independent con-
trol actuators than degrees of freedom (DOF). Aircraft,
spacecraft, helicopters, and underwater vehicles are all
examples of underactuated systems [3]. Underactuated
mechanical system control has received much attention
over the last few years due to the challenges of theory and
practice [4].
The ball-and-beam system (BABS) is a typical nonlin-
ear system that is widely used to assess the performance
of new controller design methods. There are two main
difficulties in controller design for such a BABS. The
first difficulty is that nonlinearities degrade control per-
formance [3]. A linear approximate model is widely u sed
to enable the application of linear design methods such as
PID, pole assignment, and optimal control. These meth-
ods do not handle nonlinearities precisely, however, and
thus may noise problems in control precision and small
working area or even in system stability. Intelligent con-
trol methods such as fuzzy control and genetic algorithms
are proposed to solve these problems [5]. Although these
methods do not require exact model information, their
control accuracy and plant model depend strongly on mo-
tion experiments and a high amount of parameter tuning.
Due to computational complexity, other methods such as
Lyapunov and exact linearization have been little used [6].
There is also friction on the motion surface, so the BABS
is not rigidly smooth. While this point is rarely mentioned
in reports, it adversely affects control p erformance. The
second difficulty in controller design is its underactuation.
Commonly u sed control strategies such as the state feed-
back cannot be applied due to singularity in the BABS.
The main contribution of this paper is the stabiliza-
tion of the BABS by using an improved conventional
SOSM controller. The exact model and nonlinear fric-
tion in the system are considered to obtain better control
performance. The singularity problem is solved by us-
ing a virtual controller that is realized by using improved
SOSM controller, thereby overcoming the influence of
uncertainty.
This paper is organized as follows. Section 2 gives the
BABS model and transforms it into a cascade normal form
to make controller design easier. Section 3 presents a vir-
tual controller to a second-order subsystem, defines the
Vol.18 No.2, 2014 Journal of Advanced Computational Intelligence 121
and Intelligent Informatics