Professional Documents
Culture Documents
GRADUATION THESIS
(ADVANCED PROGRAM)
GRADUATION THESIS
Topic:
BACKSTEPPING SLIDING MODE CONTROLLER FOR
NON-LINEAR ACTIVE SUSPENSION SYSTEM
1. Contents:
Chapter 1: Introduction
• Overview of active suspension systems
• Quarter-, half- and full-car models
Chapter 2: Backstepping sliding mode controller for quarter-car model
• Mathematical modeling
• Control objectives
• Zero dynamics
• Simulation results
Chapter 3: Backstepping sliding mode controller for half-car model
• Mathematical modeling
• Control objectives
• Zero dynamics and performance constraints
• Simulation results
Chapter 4: Backstepping sliding mode controller for full-car model
• Mathematical modeling
• Control objectives
• Simulation results
2. Instructor: Assoc. Prof. Nguyen Tung Lam
3. Topic assignment date:..........................................................................
4. Topic completion date:..........................................................................
Date......................................
HEAD OF DEPARTMENT INSTRUCTOR
STUDENT
ACKNOWLEDGEMENT
TABLE OF FIGURES.......................................................................................................ii
LIST OF TABLES............................................................................................................iii
ABSTRACT.......................................................................................................................iv
CHAPTER 1. INTRODUCTION.....................................................................................1
2.4 Conclusion...............................................................................................................19
References.........................................................................................................................39
LIST OF ABBREVIATIONS
Abbreviations Meaning
DOF Degree Of Freedom
BSMC Backstepping Sliding Mode Control
PID Proportional – Integral - Derivative
i
TABLE OF FIGURES
ii
LIST OF TABLES
iii
ABSTRACT
iv
CHAPTER 1. INTRODUCTION
1.1.1 Overview
Suspension system transmits all the forces from the interactions between the
vehicle body to the road. Therefore, the system is the main element which determines
ride comfort, road holding, and ride safety. The quality of the ride is also significantly
influenced by the performance of the suspension system, as it does not only make
impact on the driver’s comfort but also prevent physical fatigue and reduce the rate of
fatality in traffic accidents [1][2]. A basic model of a car suspension system is
illustrated in Figure 1.
1
Car suspension systems can be categorized into 3 types: Passive, semi-active and
active suspension systems. Passive suspension system consists of springs and dampers
positioned between the body of the vehicle and the wheel-axle assembly. Passive
suspensions are preferred when it comes to high reliability, simple mechanism and
implementation. However, they show weak performances in improving road comfort
and road holding under different road conditions and conflicting criteria. The reason
for this property is the invariant characteristics of the spring stiffness and the damping
coefficient [3][4].
Semi-active suspension systems provide variable springs and dampers, which
enables the spring stiffness and the damping coefficient to be adjusted within a certain
range. Due to there low energy consumption and high reliability, they are used in a
vast array of production vehicles [5][6]. However, the produced damping or spring
forces are restricted by passivity constraints, i.e., they can only react to the relative
motion of the damper and dissipate energy passively. Although they present a
considerable improvement in performance compared to passive systems, semi-active
systems have limited capability in improve ride comfort.
Different from the aforementioned two kinds of suspension systems, active
suspension systems require a power supply and can produce forces independent of the
relative suspension motion. Despite of their advantages, active suspension systems
have not been applied in any production vehicles, due to their energy requirements as
well as weight and packaging aspects. But there is no doubt that active suspensions
will be integrated in the trend of the future vehicle suspension design [7][8].
The success of an active suspension system is achieved through the design process
comprised of two stages. The first stage is to build a control-oriented dynamic model
of the vehicle active suspension. The second stage is to design a suitable control
strategy, which has a significant impact on ride comfort and ride safety.
2
of suspension systems: 2-DOF quarter-car model, 4-DOF half-car model, 7-DOF full-
car model [9].
101\*
MERGEFORMAT (.)
3
where is the quarter-car body mass; u is the unsprung mass (tire, wheel,
brake calliper, suspension links, etc.), are the displacements of the sprung and
are the stiffness of the spring and the tire, respectively; and are the
damping coefficients of the sprung absorber and the tire, respectively. is the actuator
input force.
In 01, the damper force and spring force are considered linear, which implies that
the quarter-car dynamic model behaves as a linear system. The linear model facilitates
performance analysis, controllability and so on. However, the actual damping
coefficient is different during the extension and compression movements, and the
spring stiffness coefficient is also non-linear in nature. Non-linear model and control
strategy will be discussed later in the following chapters.
4
and are the mass of the vehicle body and the moment of inertia for the ptich motion,
respectively; stand for the front and rear unsprung masses, respectively.
denote the forces produced by the front and rear springs and dampers,
respectively; and are the elasticity and damping forces of the front and
rear tires. and are friction forces of suspension components. is the vertical
displacement, is the pitch angle, are the displacements of the unpsrung masses
and are the road inputs to the corresponding wheel. are the distances of the
suspensions to the center of gravity of the vehicle body, and are the control
202\* MERGEFORMAT
(.)
The forces generated by the springs and dampers can be modeled as both linear and
non-linear form.
5
1.1.2.3 Full-car Model
as depicted in Figure 5. In Figure 4, stand for the mass of the vehicle body,
moment of inertia for the roll and pitch motions, respectively. are the
unsprung masses of front left, front right, rear left, and rear right, respectively. and
are the forces produced by the spring and damper, respectively, and is the
stiffness coefficient of the tire. For the vehicle body, , and represent the heave,
pitch and roll motions, respectively. is the unsprung mass displacement and is
the road input to the respective wheel. The actuators are placed parallel to the
6
suspension springs and dampers, and their output forces are denoted by .
represent the the distances of the suspension to the center of mass of the vehicle body.
is the velocity of the vehicle in x-direction.
Figure 5. (a) Non-linear spring and (b) piece-wise linear damper models.
The dynamic equations of the full-car model are obtained as follows, based on an
assumption that the pitch and roll angles are small.
303\*
MERGEFORMAT (.)
linear term and is the stiffness coefficient of the cubic term. Here, is the
damping coefficient for the extension movement, is the damping coefficient for the
7
compression movement. , stand for the suspension deflections, which are
defined as:
404\* MERGEFORMAT
(.)
505\* MERGEFORMAT
(.)
606\* MERGEFORMAT
(.)
707\* MERGEFORMAT
(.)
808\*
MERGEFORMAT (.)
8
where A represents the bump height, L is the bump length, and v is the velocity of
the passing vehicle.
9
problems of actuator input delay, sampled data, and fault accommodation for
unknown actuator failures.
10
Figure 6. (a) Ideal trajectory and (b) Non-ideal trajectory of system in SMC
To enhance the performance of SMC, backstepping technique is implemented. The
basic idea of backstepping design is that a complex non-linear system is decomposed
into the subsystems, and the degree of each subsystem doesn’t exceed that of the
whole system. Accordingly, the Lyapunov function and virtual control signals are
designed respectively, and the whole system is obtained through “backstepping”. Thus
the control rule is designed thoroughly. The backstepping method is called as back-
deduce method, and the desired dynamic indexes are satisfied. The backstepping
method and the sliding mode control are integrated to designed a backstepping-sliding-
mode controller which realizes the robust control for uncertain systems. It combines
the immunity to uncertainties of backstepping technique and the robustness of sliding
mode control to external disturbances. Utilizing this property, many researchers have
implemented this control algorithm for a wide range of non-linear objects which
contain uncertainties in the model and are susceptible to disturbances [15-17]. The
non-linear nature of the active suspension system makes it highly suitable for applying
backstepping sliding mode approach to achieve great performance.
Suppose the plant is a non-linear system as below:
909\* MERGEFORMAT
(.)
11
where are non-linear functions, is the sum of the
12012\*
MERGEFORMAT (.)
where is the sliding variable. It is obtained that
If , then .
Select a Lyapunov function as
Because , we obtain
12
To have , a controller is designed as below
14014\*
MERGEFORMAT (.)
Therefore, as .
13
CHAPTER 2. Backstepping Sliding Mode Controller for Active
as in Figure 2 . 2-DOF are vertical motions of the sprung mass and the unsprung
configuration and an actuator to transform this system into an active one. Moreover,
unsprung mass contacts with the road surface through a tire, which is modeled as a
spring-damper. No disturbance and uncertainty is concerned in the model [9].
14
Figure 7. Active suspension system in quarter-car model
Dynamic equations of the quarter-car model:
16116\* MERGEFORMAT
(.)
in which are sprung and unsprung mass, respectively. are the forces
from the spring and damper, respectively. are the forces from the spring and
damper of the tire. are vertical accelerations of the sprung and unsprung masses,
defined as follows:
17117\* MERGEFORMAT
(.)
18118\* MERGEFORMAT
(.)
19119\*
MERGEFORMAT (.)
where are the stiffness coefficient of the linear terms and the cubic terms,
respectively; are the damping coefficient for the extension movement and the
15
20120\*
MERGEFORMAT (.)
21121\*
MERGEFORMAT (.)
16
vibrations, the standard deviation is considered for the analysis of the
26126\* MERGEFORMAT
(.)
Define
17
28128\* MERGEFORMAT (.)
Hence,
29129\*
MERGEFORMAT (.)
We have , then
31131\*
MERGEFORMAT (.)
where . Hence,
Then, as .
18
In order to reduce chattering phenomenon, is replaced by with
32132\*
MERGEFORMAT (.)
19
Bump length 5m
Actuator saturation 5000 N
Table 2. Controller parameters for quarter-car model
Controller BSMC PID
Parameters
Value 5 5 1 10 1 10000
Simulation results in MATLAB/Simulink are presented from Figure 8 to Figure 12.
Road input disturbance is given by 08 in 1.1.3 and influences the system at the third
second of simulation.
20
Figure 10. Control input
Figure 12. Ratio between dynamic tire load and static tire load
Generally, active suspension systems (BSMC and PID) bring better ride comfort
than passive suspension system does. Figure 8 and Figure 9 depict an excellent
21
performance of BSMC, as the sprung mass moves at remarkably small amplitude and
acceleration compared to PID controller and passive suspension system. BSMC
guarantees great ride comfort, good road holding and maintains ride safety at high
level. Compared to PID, BSMC produces a lower vertical displacement of sprung
mass and a smaller amplitude of vertical acceleration. Control input, suspension travel
and dynamic tire load in every control scheme are kept under their limits, as shown in
Figure 10, Figure 11 and Figure 12.
2.4 Conclusion
3312Equation Chapter (Next) Section 1In this chapter, ride comfort of vehicle’s
occupants is discussed and improved by using a BSMC for active suspension system
of quarter-car model. In general, passive suspension system is not able to provide
comfort to the passengers under uneven road condition. Comfort level when driving of
people travelling in the vehicle is significantly improved by reducing vertical
acceleration of the sprung mass. Simultaneously, hard constraints of the system are
fully satisfied: Actuator saturation, road holding based on dynamic and static tire loads
and maximum suspension travel. A comparison among active suspension systems
using BSMC, conventional PID, and passive suspension system is conducted.
Numerical simulations are performed and show that active suspension with BSMC
brings superior comfort, and satisfies all requirements of the system.
22
CHAPTER 3. Backstepping Sliding Mode Controller for Active
If the left and right side of the car are symmetrical, the suspension model can be
simplified as a half-car model. Compared to the quarter-car model, half-car suspension
systems can reflect both vertical and pitch motions. This property makes half-car
model appropriate for testing new control algorithm. This chapter will analyze the
half-car model and design a BSMC for this model.
23
34234\*
MERGEFORMAT (.)
in which
The forces produced by the non-linear stiffening spring, the piece-wise linear
damper and the tires:
35235\*
MERGEFORMAT (.)
36236\*
MERGEFORMAT (.)
37237\*
MERGEFORMAT (.)
38238\*
MERGEFORMAT (.)
39239\*
MERGEFORMAT (.)
Let .
24
40240\*
MERGEFORMAT (.)
41241\*
MERGEFORMAT (.)
with satisfiy:
− Suspension space limits: Because of mechanical structure, the
suspension spaces should not exceed the allowable maximums:
42242\* MERGEFORMAT
(.)
− Actuator saturation: All actuators of physical devices are subject to
amplitude saturation:
25
3.2 Synthesis of BSMC for Half-car Model
system. By determining the control signals for the heave motion and for pitch
motion of the system, the real control inputs can be calculated. First, control
signal is designed so that tracking error converges to 0. Dynamic equations for heave
motion:
with .
26
In order to obtain , let , i.e.
47247\*
MERGEFORMAT (.)
function
Because , we have
49249\*
MERGEFORMAT (.)
where .
Hence,
As a result, as .
50250\*
MERGEFORMAT (.)
27
which leads to
52252\*
MERGEFORMAT (.)
53253\*
MERGEFORMAT (.)
Similarly, the control signal for pitch motion of the system is designed as:
54254\*
MERGEFORMAT (.)
56256\* MERGEFORMAT
(.)
28
3.2.2 Zero dynamics
The backstepping sliding mode controller yields to a 4th order error dynamic, while
the original system is 8th order. Therefore, the zero dynamics consists of 4 states. Let
, we have
where
derivative of V:
29
It is worth noting that matrix A has eigenvalues with negative real parts. Hence,
We can properly choose matrices and tuning values so that there exists
where with
259 suggests that the Lyapunov function is bounded by
60260\* MERGEFORMAT
(.)
30
3.2.3 Performance Constraints
We know that all signals are bounded within known ranges, and the bounds of
dynamic tire loads can be estimated as
61261\*
MERGEFORMAT (.)
62262\*
MERGEFORMAT (.)
If we choose properly the initial values and tuning parameters, then these below
can always be obtained:
63263\*
MERGEFORMAT (.)
64264\*
MERGEFORMAT (.)
65265\*
MERGEFORMAT (.)
31
If the initial values and tuning gains are properly adjusted, then the
67267\*
MERGEFORMAT (.)
that
Road input disturbance is given by 08 in 1.1.3 and influences the system at the
32
Unsprung mass 100 kg
33
Figure 14. Vertical displacement in different control schemes
34
Figure 16. Pitch angle in different control schemes
35
Figure 18. Vertical acceleration in different control scheme
36
Figure 20. Pitch acceleration in different control schemes
3.4 Conclusion
In this chapter, half-car model is controlled by a BSMC. The simulation results
depicts great control quality, as it is illustrated by numerical simulation and
comparison to other control strategies. Short transient time and low displacement
amplitude are important advantages of using BSMC for half-car model. In the next
chapter, BSMC is applied to 7-DOF full-car model.
6813Equation Chapter (Next) Section 1
37
CHAPTER 4. Backstepping Sliding Mode Controller for Active
38
69369\*
MERGEFORMAT (.)
where
71371\* MERGEFORMAT
(.)
72372\* MERGEFORMAT
(.)
75375\*
MERGEFORMAT (.)
76376\*
MERGEFORMAT (.)
77377\*
MERGEFORMAT (.)
39
78378\*
MERGEFORMAT (.)
Here, are the stiffness of the linear term and is the stiffness coefficient of the
cubic term of the non-linear spring. Here, is the damping coefficient for the
40
79379\*
MERGEFORMAT (.)
80380\* MERGEFORMAT
(.)
where
41
81381\*
MERGEFORMAT (.)
Let
84384\*
MERGEFORMAT (.)
where is the sliding variable. Therefore
We have . It leads to
42
To have , choose control law
86386\*
MERGEFORMAT (.)
where . As a result
Therefore, as .
87387\*
MERGEFORMAT (.)
Similar to the synthesis of BSMC for heave motion, BSMCs for roll and pitch
motion are obtained:
− For pitch motion:
88388\*
MERGEFORMAT (.)
where and
43
89389\
* MERGEFORMAT (.)
− For roll motion:
90390\*
MERGEFORMAT (.)
where and
91391\
* MERGEFORMAT (.)
44
Distance from the suspensions to
the center of gravity of the vehicle
Parameter
Value
45
46
47
RMSVA
PSS
BSMC
References
48
[7] H. Li, H. Liu, H. Gao et al., “Reliable fuzzy control for active suspension
systems with actuator delay and fault,” in IEEE Trans. Fuzzy Syst. 20(2), 342–357,
2012.
[8] G. Priyandoko, M. Mailah, H. Jamaluddin, “Vehicle active suspension system
using skyhook adaptive neuro active force control,” in Mech. Syst. Signal Process.
23(3), 855–868, 2009.
[9] W. Sun, H. Gao, and P. Shi, “Advanced Control for Vehicle Active Suspension
Systems,” in vol. 204, no. 2. Cham: Springer International Publishing, 2020.
[10] H. Pang, X. Zhang, J. Chen, and K. Liu, “Design of a coordinated adaptive
backstepping tracking control for nonlinear uncertain active suspension system,” in
Appl. Math. Model., vol. 76, pp. 479–494, 2019, doi: 10.1016/j.apm.2019.06.030.
[11] E. Lavretsky and T. E. Gibson, “Projection Operator in Adaptive Systems,”
2011, [Online]. Available: http://arxiv.org/abs/1112.4232.
[12] B. Yao, J. Reedy, and G. T. C. Chiu, “Adaptive Robust Motion Control Of
Single-Rod Hydraulic Actuators: Theory and Experiments,” no. June, pp. 759–763,
1999.
[13] Edwards C, Spurgeon S., “Sliding Mode Control: Theory and Applications,”
London: Taylor and Francis, 1998.
[14] Cao, X. Song, M. Ahmadian, “Editors perspectives: road vehicle suspension
design, dynamics, and control,” in Veh. Syst. Dyn. 49(1–2), 3–28, 2011.
[15] V. T. Dang et al., “Adaptive Control for Multi-Shaft with Web Materials
Linkage Systems,” Inventions, vol. 6, no. 4, p. 76, Oct. 2021, doi:
10.3390/inventions6040076.
[16] D. T. Le, V. T. Dang, B. H. N. Dinh, H. P. Vu, V. P. Pham, and T. L. Nguyen,
“Disturbance Observer-Based Speed Control of Interior Permanent Magnet
Synchronous Motors for Electric Vehicles,” in Inflammation, 2022, pp. 244–259.
[17] D. T. Le, D. T. Nguyen, N. D. Le, and T. L. Nguyen, “Traction control based
on wheel slip tracking of a quarter-vehicle model with high-gain observers,” Int. J.
Dyn. Control, Oct. 2021, doi: 10.1007/s40435-021-00881-6.
49