Send Orders for Reprints to reprints@benthamscience.ae
826 The Open Automation and Control Systems Journal, 2015, 7, 826-834
1874-4443/15 2015 Bentham Open
Open Access
Design and Realization of Position Controller Based on Characteristic
Model for Servo Systems with Large Inertia Ratio
Zhihong Wang
1,*
, Wei Chen
2
, Yifei Wu
1
, Xiang Wang
1
, Jian Guo
1
and Qingwei Chen
1
1
School of Automation, Nanjing University of Science and Technology, Nanjing 210094, China
2
Electronic and Mechanical Technology Department of the Nanjing Marine Radar Institute, Nanjing 210003, China
Abstract: An adaptive sliding mode controller (ASMC) based on characteristic model is designed to overcome the detri-
mental effect of large inertia ratio and large-range varying inertia in high accuracy servo systems. The servo system dis-
crete characteristic model is established and adopted for the controller designed instead of using the traditional mechanism
model. The recursive least square (RLS) algorithm is used to identify time-varying parameters in characteristic model.
The position controller is constituted by an adaptive equivalent controller based on identification parameters and an im-
proved sliding mode controller, and the stability of the closed-loop system is analyzed. The experimental results show that
the proposed controller can adapt to large-range varying inertia, and improve the dynamic performance and steady-state
precision of servo systems.
Keywords: Large inertia ratio, Servo systems, Characteristic model, Adaptive sliding mode control.
1. INTRODUCTION
Servo systems have been widely used in various applica-
tions, such as weapon systems, radars, machine tools, robots
and astronomical telescopes. With the rapid development of
national defense and national economy, the requests for high
precision, good dynamic performance and good adaptability
have been proposed to servo systems. For instance, a sup-
pressing multiple launch rocket system cannot be regarded as
a rigid body, because the long barrel structure makes it flexi-
ble. In the launching process, the load inertia and torque var-
ies in a large range due to varying bomb load. What’s more,
the nonlinearities such as uncertain friction are present in the
system which makes it more difficult for traditional control
algorithms to meet the control requirements.
Considering the large range variation of system parame-
ters and disturbance torque, high speed and high precision
control algorithms with good adaptability and disturbance
rejection ability are required. Existing control methods in-
cludes sliding mode control [1-3], Kalman filter prediction
control [4], neural networks control [5], wavelet based slid-
ing mode control [6, 7], disturbance observer [8, 9] etc. In
[2], an adaptive backstepping sliding mode controller was
proposed for pitch position control system of the rocket
launcher, backstepping control was adopted to guarantee the
stability of the closed-loop tracking system, and sliding
mode control was adopted to restrain the parameter perturba-
tion and external disturbances. In [5], a sliding mode adap-
tive controller was designed based on neural network, and
*Address correspondence to this author at the School of Automation, Nan-
jing University of Science and Technology, Nanjing 210094, China;
E-mail: njust.wang@gmail.com
neural network was applied to replace the switching function
in sliding mode control. In [6], fuzzy wavelet neural network
was used to approximate the equivalent control variable and
a terminal sliding mode position controller was designed for
high power weapon servo systems. In [10], an adaptive con-
troller based on extended state observer and inertia estima-
tion was developed for servo systems with variable inertia.
Existing methods can achieve good control performance
when system parameters or load inertia varies within a small
range, but there are still many deficiencies such as depend-
ence of precise mathematical model, multiple adjustable pa-
rameters, complex debugging process, difficult determina-
tion of parameter range and slow convergence of the adap-
tive parameters. It is also difficult to achieve good control
performance when the system parameters vary in a large
range.
In our research, the characteristic model is adopted to de-
scribe the servo systems instead of the traditional mechanism
model [10] or other modeling methods of different fields
[11-13], which are fairly complicated or not suitable for ser-
vo systems compared with the characteristic model. Charac-
teristic modeling is a new modeling method proposed by Wu
[14, 15]. The system satisfying certain conditions can be
equivalent with a second-order linear time-variant system
which makes it easier to design the controller, and the pa-
rameters can be determined beforehand within a fairly small
range, which is beneficial to fast parameter convergence
when estimating the parameters online with the RLS algo-
rithm. The nonlinear components and system parameter
changes are also considered in the characteristic model
which makes the designed controller with better performance
and robustness. In sliding mode control, the system states are
forced to run on the sliding-mode surface, and it has good