04/12/2020

System Dynamics and Control 1 Design via State Space System Dynamics and Control 2 Design via State Space

Learning Outcome
After completing this chapter, the student will be able to
• Design a state-feedback controller using pole placement for
systems represented in phase-variable form to meet transient
response specifications
• Determine if a system is controllable
• Design a state-feedback controller using pole placement for
systems not represented in phase-variable form to meet
transient response specifications
• Design a state-feedback observer using pole placement for
systems represented in observer canonical form
• Determine if a system is observable
• Design a state-feedback observer using pole placement for
systems not represented in observer canonical form
• Design steady-state error characteristics for systems
represented in state space
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 3 Design via State Space System Dynamics and Control 4 Design via State Space

§1.Introduction §1.Introduction
- Frequency domain methods of design do not allow to specify all - Frequency domain methods do allow the specification of
poles in systems of order higher than 2 because they do not closed-loop zero locations, which time domain methods do not
allow for a sufficient number of unknown parameters to place allow through placement of the lead compensator zero
all of the closed-loop poles uniquely This is a disadvantage of state-space methods, since the
→ State-space methods solve this problem by introducing location of the zero does affect the transient response
(1) other adjustable parameters, and Also, a state-space design may prove to be very sensitive to
(2) the technique for finding these parameter values, so that parameter changes
we can properly place all poles of the closed-loop system

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 5 Design via State Space System Dynamics and Control 6 Design via State Space

§1.Introduction §2.Controller Design

- There is a wide range of computational support for state-space - An 𝑛th-order feedback control system has an 𝑛th-order closed-
methods; many software packages support the matrix algebra loop characteristic equation of the form
required by the design process 1𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0 = 0 (12.1)
However, as mentioned before, the advantages of computer Since the coefficient of the highest power of 𝑠 is unity, there are
support are balanced by the loss of graphic insight into a 𝑛 coefficients whose values determine the system’s closed-loop
design problem that the frequency domain methods yield pole locations. Thus, if we can introduce 𝑛 adjustable
- This chapter considers only an introduction to state-space parameters into the system and relate them to the coefficients
design only to linear systems in Eq. (12.1), all of the poles of the closed-loop system can be
set to any desired location

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

1
 04/12/2020

System Dynamics and Control 7 Design via State Space System Dynamics and Control 8 Design via State Space

§2.Controller Design §2.Controller Design

Topology for Pole Placement - A plant signal-flow graph in phase-variable (controller canonical) form
- Consider the closed-loop system represented in state space 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 𝑐3 (12.2.a)
𝑐2
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 𝑢 𝒙ሶ 𝒙 𝑦
(12.2.a) 𝑦 = 𝑪𝒙 (12.2.b)
1 1/𝑠 1/𝑠 1/𝑠 𝑐1
𝑦 = 𝑪𝒙 𝑩 𝑪 (12.2.b) 𝑢
−𝑎2 𝑥3 𝑥2 𝑥1
𝑦
−𝑎1
𝑨 −𝑎0

State space representation of a plant with no feedback Phase variable representation for plant with no feedback
𝑐3
𝑟 𝑢 𝒙ሶ 𝒙 𝑦 𝑐2
𝑩 𝑪
1 1 1/𝑠 1/𝑠 1/𝑠 𝑐1
𝑟 𝑦
𝑨 𝑢 −𝑎2 𝑥3 𝑥2 𝑥1
−𝑘3 −𝑎1
−𝑘2 −𝑎0
𝑲 −𝑘1
State space representation of a plant with state variable feedback Phase variable representation for plant with state variable feedback

𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝑨𝒙 + 𝑩 −𝑲𝒙 + 𝑟 = 𝑨 − 𝑩𝑲 𝑥 + 𝑩𝑟 (12.3.a) 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝑨𝒙 + 𝑩 −𝑲𝒙 + 𝑟 = 𝑨 − 𝑩𝑲 𝑥 + 𝑩𝑟 (12.3.a)

𝑦 = 𝑪𝒙 (12.3.b) 𝑦 = 𝑪𝒙 (12.3.b)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 9 Design via State Space System Dynamics and Control 10 Design via State Space

§2.Controller Design §2.Controller Design

Pole Placement for Plants in Phase-Variable Form Pole Placement for Plants in Phase-Variable Form
To apply pole-placement methodology to plants represented in 1.Represent the plant in phase-variable form
𝑐3
phase-variable form 𝑐2
1.Represent the plant in phase-variable form 1 1/𝑠 1/𝑠 1/𝑠 𝑐1
𝑢 −𝑎2 𝑦
2.Feed back each phase variable to the input through a gain 𝑘𝑖 −𝑎1
𝑥2 𝑥1

3.Find the characteristic equation for the closed-loop system −𝑎0

represented in Step 2 The phase-variable representation of the plant is given by

4.Decide upon all closed-loop pole locations and determine an 0 1 0 ⋮ 0 0
equivalent characteristic equation 0 0 1 ⋮ 0 0
𝑨= , 𝑩 = ,𝑪 = 𝑐1 𝑐2 ⋯ 𝑐𝑛
5.Equate like coefficients of the characteristic equations from ⋮ ⋮ ⋮ ⋱ ⋮ ⋮
Steps 3 and 4 and solve for 𝑘𝑖 −𝑎0 −𝑎1 −𝑎2 ⋯ −𝑎𝑛−1 1 (12.4)
The characteristic equation of the plant
𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0 = 0 (12.5)

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 11 Design via State Space System Dynamics and Control 12 Design via State Space

§2.Controller Design §2.Controller Design

2.Feed back each phase variable to the input through a gain 𝑘𝑖 3.Find the characteristic equation for the closed-loop system
Form the closed-loop system by feeding back each state represented in Step 2
variable to 𝑢 𝑐3 det 𝑠𝑰 − 𝑨 − 𝑩𝑲 = 0
𝑐 2
→ 𝑠 𝑛 + 𝑎𝑛−1 + 𝑘𝑛 𝑠 𝑛−1 + ⋯ + 𝑎1 + 𝑘2 𝑠 + 𝑎0 + 𝑘1 = 0
1 1 1/𝑠 1/𝑠 1/𝑠 𝑐1
𝑟 𝑦
𝑢 −𝑎2 𝑥2 𝑥1 (12.9)
−𝑘3 −𝑎1
−𝑘2 −𝑎0 (12.9) can be derived from (12.5) by adding the appropriate 𝑘𝑖
−𝑘1
𝑢 = −𝑲𝒙 (12.6) to each coefficient
where 𝑘𝑖 : the phase variables’ feedback gains 4.Decide upon all closed-loop pole locations and determine an
equivalent characteristic equation
The system matrix, 𝑨 − 𝑩𝑲, for the closed-loop system
0 1 0 ⋮ 0 𝑠 𝑛 + 𝑑𝑛−1 𝑠 𝑛−1 + 𝑑𝑛−2 𝑠 𝑛−2 + ⋯ + 𝑑2 𝑠 2 + 𝑑1 𝑠 + 𝑑0 = 0
0 0 1 ⋮ 0 (12.10)
𝑨−𝑩𝑲 = (12.8)
⋮ ⋮ ⋮ ⋱ ⋮
−(𝑎0 +𝑘1) −(𝑎1 +𝑘2) −(𝑎2 +𝑘3) ⋯ −(𝑎𝑛−1 +𝑘𝑛) 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0 = 0 (12.5)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

2
 04/12/2020

System Dynamics and Control 13 Design via State Space System Dynamics and Control 14 Design via State Space

§2.Controller Design §2.Controller Design

- Ex.12.1 Controller Design for Phase-Variable Form
5.Equate like coefficients of the characteristic equations from Design the phase-variable feedback gains to yield %𝑂𝑆 = 9.5%
Steps 3 and 4 and solve for 𝑘𝑖 and 𝑇𝑠 = 0.74𝑠 20(𝑠 + 5)
𝐺 𝑠 =
Equating Eqs. (12.9) and (12.10) to obtain Solution 𝑠(𝑠 + 1)(𝑠 + 4)
𝑑𝑖 = 𝑎𝑖 + 𝑘𝑖+1 𝑖 = 0,1,2, … ,𝑛 − 1 Represent the plant in phase-variable form
𝑘𝑖+1 = 𝑑𝑖 − 𝑎𝑖 20(𝑠 + 5) 1
𝐺 𝑠 = = × (20𝑠 + 100)
Note: • For systems represented in phase-variable form, the 𝑠(𝑠 + 1)(𝑠 + 4) 𝑠 3 + 5𝑠 2 + 4𝑠 + 0
20
numerator polynomial is formed from the coefficients of
the output coupling matrix, 𝑪 𝑢
1 1/𝑠
−5
1/𝑠 1/𝑠 100
𝑦
𝑥3 𝑥2 𝑥1
• The plan and closed-loop system are both in phase- −4

variable form and have the same output coupling matrix 0 1 0 0

→ the numerators of their transfer functions are the same 𝒙ሶ = 0 0 1 𝒙 + 0 𝑟, 𝑦 = 100 20 0 𝒙
𝑠 𝑛 + 𝑎𝑛−1 + 𝑘𝑛 𝑠 𝑛−1 + ⋯ + 𝑎1 + 𝑘2 𝑠 + 𝑎0 + 𝑘1 = 0 (12.9) 0 −4 −5 1
𝑠 𝑛 + 𝑑𝑛−1 𝑠 𝑛−1 + 𝑑𝑛−2 𝑠 𝑛−2 + ⋯ + 𝑑2 𝑠 2 + 𝑑1 𝑠 + 𝑑0 = 0 (12.10) → 𝑠 3 + 5𝑠 2 + 4𝑠 + 0 = 0
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 15 Design via State Space System Dynamics and Control 16 Design via State Space

§2.Controller Design §2.Controller Design

Feed back each phase variable to the input through a gain 𝑘𝑖 Decide the closed-loop pole locations and determine an
20
𝑢 = −𝑲𝒙 equivalent characteristic equation
1 1 1/𝑠 1/𝑠 1/𝑠 100
𝑟
𝑢 −4 𝑥3 𝑥2 𝑥1
𝑦 The second-order system with the desired performances
−𝑘3 −5 𝑙𝑛(%𝑂𝑆/100) 𝑙𝑛(9.5/100)
−𝑘2
−𝑘1
ζ=− =− = 0.5996
𝜋 2 + 𝑙𝑛2(%𝑂𝑆/100) 𝜋 2 + 𝑙𝑛2 (9.5/100)
Find the characteristic equation for the closed-loop system 4 4
𝜔𝑛 = = = 9.0147
The closed-loop system’s state equations ζ 𝑇𝑠 0.5996 × 0.74
2
𝜔𝑛 81.2648
0 1 0 0 →𝐺 𝑠 = 2 =
𝒙ሶ = 0 0 1 𝒙 + 0 𝑟, 𝑦 = 100 20 0 𝒙 𝑠 + 2ζ𝜔𝑛 𝑠 + 𝜔𝑛2 𝑠 + 5.4 − 𝑗7.2 [(𝑠 + 5.4 + 𝑗7.2 ]
−𝑘1 −(5 + 𝑘2) −(4 + 𝑘3) 1 3rd-order system → select another closed-loop pole: 𝑝3 = −5.1
The closed-loop system’s characteristic equation The desired characteristic equation
det 𝑠𝑰 − 𝑨 − 𝑩𝑲 = 0 𝑠 + 5.4 − 𝑗7.2 𝑠 + 5.4 + 𝑗7.2 (𝑠 + 5.1) = 0
→ 𝑠 3 + 4 + 𝑘3 𝑠 2 + 5 + 𝑘2 𝑠 + 𝑘1 = 0 (12.16) → 𝑠 3 + 15.9𝑠 2 + 136.08𝑠 + 413.1 = 0 (12.17)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 17 Design via State Space System Dynamics and Control 18 Design via State Space

§2.Controller Design §2.Controller Design

Equate like coefficients of the characteristic equations from
Steps 3 and 4 and solve for 𝑘𝑖
Equating Eqs. (12.16) and (12.17) to obtain
𝑘1 = 413.1, 𝑘2 = 132.08, 𝑘3 = 10.9
The state-space representation of the closed-loop system
0 1 0 0
𝒙ሶ = 0 0 1 𝒙+ 0 𝑟 (12.19.a)
−413.1 −136.08 −15.9 1
𝑦 = 100 20 0 𝒙 (12.19.b)
The closed-loop transfer function
From the simulation results: %𝑂𝑆 = 10.1%, 𝑇𝑠 = 0.664𝑠
20(𝑠 + 5)
𝑇 𝑠 = 3 The steady-state response approaches 0.242 instead of unity,
𝑠 + 15.9𝑠 2 + 136.08𝑠 + 413.1
there is a large steady-state error → Design techniques to
𝑠 3 + 5 + 𝑘3 𝑠 2 + 4 + 𝑘2 𝑠 + 𝑘1 = 0 (12.16) reduce this error are discussed in Section 12.8
𝑠 3 + 15.9𝑠 2 + 136.08𝑠 + 413.1 = 0 (12.17)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

3
 04/12/2020

System Dynamics and Control 19 Design via State Space System Dynamics and Control 20 Design via State Space

§2.Controller Design §2.Controller Design

Run ch12p1 in Appendix B Skill-Assessment Ex.12.1
Learn how to use MATLAB to Problem For the plant
• design a controller for phase variables using pole 100(𝑠 + 10)
𝐺 𝑠 =
placement 𝑠(𝑠 + 3)(𝑠 + 12)
• solve Ex.12.1 represented in the state space in phase-variable form by
0 1 0 0
𝒙ሶ = 0 0 1 𝒙+ 0 𝑟
0 −36 −15 1
𝑦 = 1000 100 0 𝒙
design the phase-variable feedback gains to yield
%𝑂𝑆 = 5% and 𝑇𝑝 = 0.3𝑠

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 21 Design via State Space System Dynamics and Control 22 Design via State Space

§2.Controller Design §2.Controller Design

Solution Represent the plant in phase-variable form Feed back each phase variable to the input through a gain 𝑘𝑖
1 𝑢 = −𝑲𝒙 100
𝐺 𝑠 = 3 × (100𝑠 + 1000)
𝑠 + 15𝑠 2 + 36𝑠 + 0 𝑟
1 1 1/𝑠 1/𝑠 1/𝑠 1000
𝑦
100 𝑢 −15 𝑥3 𝑥2 𝑥1
−𝑘3 −36
1 1/𝑠 1/𝑠 1/𝑠 1000 −𝑘2
𝑢 𝑦 −𝑘1
−15 𝑥3 𝑥2 𝑥1
−36
Find the characteristic equation for the closed-loop system
0 1 0 0
The closed-loop system’s state equations
𝒙ሶ = 0 0 1 𝒙+ 0 𝑟
0 −36 −15 1 0 1 0 0
𝑦 = 1000 100 0 𝒙 𝒙ሶ = 0 0 1 𝒙+ 0 𝑟
→ 𝑠 3 + 15𝑠 2 + 36𝑠 + 0 = 0 −𝑘1 −(𝑘2 + 36) −(𝑘3 + 15) 1
𝑦 = 1000 100 0 𝒙
The closed-loop system’s characteristic equation
100(𝑠+10) 𝑠 3 + 𝑘3 + 15 𝑠 2 + 𝑘2 + 36 𝑠 + 𝑘1 = 0 (1)
𝐺 𝑠 =
𝑠(𝑠+3)(𝑠+12)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 23 Design via State Space System Dynamics and Control 24 Design via State Space

§2.Controller Design §2.Controller Design

Decide the closed-loop pole locations and determine an Equate like coefficients of the characteristic equations
equivalent characteristic equation from Steps 3 and 4 and solve for 𝑘𝑖
The second-order system with the desired Equating Eqs. (1) and (2) to obtain
performances 𝑘1 = 2094, 𝑘2 = 373.1, 𝑘3 = 14.97
−𝑙𝑛(%𝑂𝑆/100) −𝑙𝑛(0.05)
ζ= = = 0.6901 The state-space representation of the closed-loop system
𝜋 2 + 𝑙𝑛2(%𝑂𝑆/100) 𝜋 2 + 𝑙𝑛2(0.05) 0 1 0 0
𝜋 𝜋 𝒙ሶ = 0 0 1 𝒙+ 0 𝑟
𝜔𝑛 = = = 14.4699
𝑇𝑝 1 − ζ 2 0.3 1 − 0.69012 −2094 −409.1 −29.97 1
𝑦 = 1000 100 0 𝒙
→ 𝑠 2 + 2ζ𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 19.97𝑠 + 209.4 = 0
The closed-loop transfer function
3rd-order system → select the third pole −10 to cancel
100(𝑠 + 10)
the zero at −10 → the desired characteristic equation 𝑇 𝑠 = 3
𝑠 + 29.97𝑠 2 + 409.1𝑠 + 2094
(𝑠 2 + 19.97𝑠 + 209.4) 𝑠 + 10 = 0
→ 𝑠 3 + 29.97𝑠 2 + 409.1𝑠 + 2094 = 0 (2) 𝑠 3 + 𝑘3 + 15 𝑠 2 + 𝑘2 + 36 𝑠 + 𝑘1 = 0 (1)
𝑠 3 + 29.97𝑠 2 + 409.1𝑠 + 2094 = 0 (2)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

4
 04/12/2020

System Dynamics and Control 25 Design via State Space System Dynamics and Control 26 Design via State Space

§2.Controller Design §2.Controller Design

TryIt 12.1
Use MATLAB, the Control 0 1 0 0
System Toolbox, and the
following statements to
𝒙ሶ = 0 0 1 𝒙+ 0 𝑟
solve for the phase-variable 0 −36 −15 1
feedback gains to place the 𝑦 = 1000 100 0 𝒙
poles of the system in Skill-
Assessment Ex.12.1 at
− 3 − 𝑗5; −3 + 𝑗5, and −10
A=[0 1 0; 0 0 1; 0 -36 -15];
B=[0;0;1];
poles=[-3+5j,-3-5j,-10];
K=acker(A,B,poles)

100(𝑠 + 10)
𝑇 𝑠 =
𝑠 3 + 29.97𝑠 2 + 409.1𝑠 + 2094
From the simulation results: %𝑂𝑆 = 5%, 𝑇𝑝 = 0.3𝑠

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 27 Design via State Space System Dynamics and Control 28 Design via State Space

§3.Controllability §3.Controllability
The system is controllable if an input to a system can take every The Controllability Matrix
state variable from a desired initial state to a desired final state An 𝑛th-order plant whose state equation is
Controllability by Inspection 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
A system with distinct eigenvalues and a diagonal system is completely controllable if the matrix
matrix is controllable if the input coupling matrix 𝐵 does not 𝑪𝑴 = [𝑩 𝑨𝑩 𝑨𝟐𝑩 ⋯ 𝑨𝒏−𝟏 𝑩]
have any rows that are zero
1/𝑠
is of rank 𝑛, where 𝑪𝑴 is called the controllability matrix
• Controllable system 1 −𝑎1 𝑥1
𝑘1

−𝑎1 0 0 1 𝑢 1 1/𝑠 𝑘2 𝑦
−𝑎2 𝑥2
𝒙ሶ = 0 −𝑎2 0 𝒙+ 1 𝑢
1/𝑠
0 0 −𝑎3 1 1 −𝑎3 𝑥3
𝑘 3

1/𝑠
• Uncontrollable system −𝑎4 𝑥1
𝑘4

−𝑎4 0 0 0 𝑢 1 1/𝑠 𝑘5 𝑦
−𝑎5 𝑥2
𝒙ሶ = 0 −𝑎5 0 𝒙+ 1 𝑢
1/𝑠
0 0 −𝑎6 1 1 −𝑎6 𝑥3
𝑘6

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 29 Design via State Space System Dynamics and Control 30 Design via State Space

§3.Controllability §3.Controllability
- Ex.12.2 Controllability via the Controllability Matrix Run ch12p2 in Appendix B
Given the system, represented by a signal-flow diagram, Learn how to use MATLAB to
1/𝑠 1 1/𝑠
determine its controllability 1 −1 𝑥2 −1 𝑥1 10 • test a system for controllability
Solution 𝑢 1 𝑦 • solve Ex.12.2
The system state equation 10
1/𝑠
−1 1 0 0 1 −2 𝑥3
𝒙ሶ = 0 −1 0 𝒙 + 1 𝑢
0 0 −2 1
There is the zero in the 𝐵 matrix, this configuration leads to
uncontrollability only if the poles are real and distinct. In this
case, the system has multiple poles at −1
0 1 −2
The controllability matrix 𝑪𝑴 = 𝑩 𝑨𝑩 𝑨𝟐 𝑩 = 1 −1 1
1 −2 4
𝑪𝑴 = −1 ≠ 0 ⟹ rank 𝑪𝑴 = 3: the system is controllable
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

5
 04/12/2020

System Dynamics and Control 31 Design via State Space System Dynamics and Control 32 Design via State Space

§3.Controllability §3.Controllability
Skill-Assessment Ex.12.2 TryIt 12.2
Problem Determine whether the system Use MATLAB, the Control −1 1 0 0
System Toolbox, and the
−1 1 2 2 following statements to solve
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 0 −1 0 𝒙 + 1 𝑢
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 0 −1 5 𝒙 + 1 𝑢 Skill-Assessment Ex.12.2 0 0 −2 1
0 3 −4 1
is controllable A=[-1 1 2; 0 -1 5; 0 3 -4];
B=[2;1;1];
Solution The controllability matrix Cm=ctrb(A,B)
2 1 1 Rank=rank(Cm)
𝑪𝑴 = 𝑩 𝑨𝑩 𝑨𝟐 𝑩 = 1 4 −9
1 −1 16
→ 𝑪𝑴 = 80 ≠ 0
rank 𝑪𝑴 = 3: the system is controllable

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 33 Design via State Space System Dynamics and Control 34 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

1st method: Matching the coefficients of det(𝑠𝐼 − (𝐴 − 𝐵𝐾)) with The state equations
the coefficients of the desired characteristic equation −2 1 0
𝒙ሶ = 𝒙+ 𝑟, 𝑦 = 10 0 𝒙
- Ex.12.3 Controller Design by Matching Coefficients −𝑘1 −(𝑘2 + 1) 1
Design state feedback for the plant represented in cascade The characteristics equation
form to yield 𝑂𝑆% = 15%, 𝑇𝑠 = 0.5𝑠 𝑠2 + 𝑘2 + 3 𝑠 + 2𝑘2 + 𝑘1 + 2 = 0 (12.32)
𝑌(𝑠) 10
Solution 𝐺 𝑠 = = The desired characteristic equation
𝑈(𝑠) (𝑠 + 1)(𝑠 + 2)
The signal-flow diagram for the plant in cascade form ln %𝑂𝑆/100 ln 15/100
𝜁 =− =− = 0.5169
• open loop 𝑢 1 1/𝑠 1 1/𝑠 10
𝑦 𝜋2 + 𝑙𝑛2 %𝑂𝑆/100 𝜋2 + 𝑙𝑛2 15/100
−1 𝑥2 −2 𝑥1
• with feedback 4 4
𝜔𝑛 = = = 15. 4769𝑟𝑎𝑑/𝑠
𝑟
1 1 1/𝑠 1 1/𝑠 10
𝑦
𝑇𝑠 𝜁 0.5 × 0.5169
𝑢 −1 𝑥2 −2 𝑥1 2 2 2
−𝑘2 → 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛 = 𝑠 + 16𝑠 + 239.5 = 0 (12.33)
−𝑘1
The state equations Equating the coefficients of Eqs. (12.32) and (12.33)
−2 1 0 𝑘2 + 3 = 16 𝑘 = 211.5
𝒙ሶ = 𝒙+ 𝑟, 𝑦 = 10 0 𝒙 ቋ→ 1
−𝑘1 −(𝑘2 + 1) 1 2𝑘2 + 𝑘1 + 2 = 239.5 𝑘2 = 13
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 35 Design via State Space System Dynamics and Control 36 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

2nd method:Transforming the system to phase variables, designing 𝒙ሶ = 𝑷−1 𝑨𝑷𝒙 + 𝑷−1 𝑩𝑢, 𝑦 = 𝑪𝑷𝒙 (12.37)
the feedback gains, and transforming the designed Controllability matrix
system back to its original state-variable representation 2
𝑪𝑴𝒙 = ቂ𝑷−1 𝑩 𝑷−1 𝑨𝑷 𝑷−1 𝑩 𝑷−1 𝑨𝑷 𝑷−1 𝑩 ⋯
- Assume a plant not represented in phase-variable form 𝑛−1
−1
𝒛ሶ = 𝑨𝒛 + 𝑩𝑢, 𝑦 = 𝑪𝒛 (12.34) 𝑷 𝑨𝑷 𝑷−1 𝑩 ]
Controllability matrix = ൣ𝑷 𝑩 𝑷 𝑨𝑷 𝑷 𝑩 𝑷 𝑨𝑷 𝑷 𝑨𝑷 𝑷−1 𝑩
−1 −1 −1 −1 −1

𝑪𝑴𝒛 = [𝑩 𝑨𝑩 𝑨𝟐 𝑩 ⋯ 𝑨𝒏−𝟏 𝑩] (12.35) ⋯ 𝑷−1 𝑨𝑷 𝑷−1 𝑨𝑷 𝑷−1 𝑨𝑷 ⋯ 𝑷−1 𝑩 ]

- Assume that the system can be transformed into the phase- = 𝑷−1 [𝑩 𝑨𝑩 𝑨𝟐 𝑩 ⋯ 𝑨𝒏−𝟏 𝑩] (12.38)
variable (𝒙) representation with the transformation Substituting Eq. (12.35) into (12.38) and solving for 𝑷
𝒛 = 𝑷𝒙 (12.36) 𝑷 = 𝑪𝑴𝒛𝑪−1 𝑴𝒙 (12.39)
Substituting this transformation into Eqs. (12.34) → the transformation matrix 𝑷 can be found from the two
𝒙ሶ = 𝑷−1 𝑨𝑷𝒙 + 𝑷−1 𝑩𝑢, 𝑦 = 𝑪𝑷𝒙 (12.37) controllability matrices

𝑪𝑴𝒛 = [𝑩 𝑨𝑩 𝑨𝟐 𝑩 ⋯ 𝑨𝒏−𝟏 𝑩] (12.35)

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

6
 04/12/2020

System Dynamics and Control 37 Design via State Space System Dynamics and Control 38 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

- Design the feedback gains, 𝑢 = −𝑲𝒙 𝒙 + 𝑟 𝒙ሶ = 𝑷−1𝑨𝑷 − 𝑷−1𝑩𝑲𝒙 𝒙 + 𝑷−1𝑩𝑟, 𝑦 = 𝑪𝑷𝒙 (12.40)
𝒙ሶ = 𝑷−1 𝑨𝑷𝒙 + 𝑷−1 𝑩𝑢 - Transform Eqs. (12.40) from phase variables back to the
= 𝑷−1𝑨𝑷𝒙 − 𝑷−1𝑩𝑲𝒙𝒙 + 𝑷−1𝑩𝑟 original representation using 𝒙 = 𝑷−1 𝒛
= 𝑷−1𝑨𝑷 − 𝑷−1𝑩𝑲𝒙 𝒙 + 𝑷−1𝑩𝑟 (12.40.a) 𝒛ሶ = 𝑨𝒛 − 𝑩𝑲𝒙𝑷−1𝒛 + 𝑩𝑟 = 𝑨 − 𝑩𝑲𝒙𝑷−1 𝒛 + 𝑩𝑟 (12.41.a)
𝑦 = 𝑪𝑷𝒙 (12.40.b) 𝑦 = 𝑪𝒛 (12.41.b)
Since this equation is in phase-variable form, the zeros of this - Comparing Eqs. (12.41) with (12.3), to find the state variable
closed-loop system are determined from the polynomial formed feedback gain, 𝑲𝒛, for the original system
from the elements of 𝑪𝑷 𝑲𝒛 = 𝑲𝒙𝑷−1 (12.42)
The TF of this closed-loop system is the same as the TF for
Eqs. (12.40), since Eqs. (12.40) and (12.41) represent the
same system. Thus, the zeros of the closed-loop TF are the
same as the zeros of the uncompensated plant, based upon
the development in Section 12.2
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝑨𝒙 + 𝑩 −𝑲𝒙 + 𝑟 = 𝑨 − 𝑩𝑲 𝑥 + 𝑩𝑟, 𝑦 = 𝑪𝒙 (12.3)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 39 Design via State Space System Dynamics and Control 40 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

- Ex.12.4 Controller Design by Transformation The controllability matrix
0 0 1
Design a state-variable feedback controller to yield a %𝑂𝑆 =
𝑪𝑴𝒛 = 𝑩𝒛 𝑨𝒛𝑩𝒛 𝑨2𝒛 𝑩𝒛 = 0 1 −3
20.8% and 𝑇𝑠 = 4𝑠 for a plant
1 −1 1
𝑠+4
𝐺 𝑠 = Since det(𝑪𝑴𝒛) = −1 ≠ 0, the system is controllable
(𝑠 + 1)(𝑠 + 2)(𝑠 + 5)
1 The characteristic equation
that is represented in cascade form det 𝑠𝑰 − 𝑨 = 𝑠 3 + 8𝑠 2 + 17𝑠 + 10 = 0
1 1/𝑠 1 1/𝑠 1 1/𝑠 4
𝑢 𝑦
−1 𝑧3 −2 𝑧2 −5 𝑧1
Phase-variable representation of the system
Solution
Using the coefficients of the above equation to write
Original system 0 1 0 0
The state equations 𝒙ሶ = 𝑨𝒙 𝒙 + 𝑩𝒙 𝑢 = 0 0 1 𝒙+ 0 𝑢
−5 1 0 0 −10 −17 −8 1
𝒛ሶ = 𝑨𝒛𝒛 + 𝑩𝒛𝑢 =
0 −2 1 𝒛 + 0 𝑢 𝑦 = 𝑪𝒙 𝒙 = 4 1 0 𝒙
0 0 −1 1 −5 1 0 0
𝑦 = 𝑪𝒛𝒛 = −1 1 0 𝒛 𝒛ሶ = 𝑨𝒛 𝒛 + 𝑩𝒛 𝑢 = 0 −2 1 𝒛 + 0 𝑢, 𝑦 = 𝑪𝒛 𝒛 = −1 1 0𝒛
0 0 −1 1
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 41 Design via State Space System Dynamics and Control 42 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

The output equation was written using the coefficients of the Design the controller using the phase-variable representation
numerator of 𝐺(𝑠), since the transfer function must be the The desired closed-loop system
same for the two representations • 𝑂𝑆 = 20.8%
The controllability matrix, 𝑪𝑴𝒙 , for the phase-variable system ቋ → the designed closed-loop system: 𝑠 2 + 2𝑠 + 5
𝑇𝑠 = 4𝑠
0 0 1
• The closed-loop zero will be at 𝑠 = −4 → choose the third
𝑪𝑴𝒙 = 𝑩𝒙 𝑨𝒙 𝑩𝒙 𝑨2𝒙 𝑩𝒙 = 0 1 −8 (12.48)
closed-loop pole to cancel the closed-loop zero
1 −8 47
Calculate the transformation matrix • The total characteristic equation of the desired closed-loop
system
The transformation matrix between the two systems
𝐷 𝑠 = 𝑠 + 4 𝑠 2 + 2𝑠 + 5
1 0 0
𝑷 = 𝑪𝑴𝒛𝑪−1 = 𝑠 3 + 6𝑠 2 + 13𝑠 + 20
𝑴𝒙 = 5 1 0 (12.49)
10 7 1 =0 (12.50)

0 0 1
𝑠+4
𝐺 𝑠 = (𝑠+1)(𝑠+2)(𝑠+5), 𝑪𝑴𝒛 = 0 1 −3
1 −1 1
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

7
 04/12/2020

System Dynamics and Control 43 Design via State Space System Dynamics and Control 44 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

The designed closed-loop system Find 𝑲𝒙
• The state equations for the phase-variable form with state- 𝑠 3 + 6𝑠 2 + 13𝑠 + 20 = 0 (12.50)
variable feedback 𝑠 3 + 8 + 𝑘3𝑥 𝑠 2 + 17 + 𝑘2𝑥 𝑠 + 10 + 𝑘1𝑥 = 0 (12.52)
0 1 0
Comparing Eq.(12.50) with (12.52)
𝒙ሶ = (𝑨𝒙 − 𝑩𝒙𝑲𝒙)𝒙 = 0 0 1 𝒙
−(10 + 𝑘1𝑥) −(17 + 𝑘2𝑥) −(8 + 𝑘3𝑥) 𝑲𝒙 = 𝑘1𝑥 𝑘2𝑥 𝑘3𝑥 = [10 − 4 − 2]
𝑦 = 𝑪𝒙 𝒙 = 4 1 0 𝒙 Transform the controller back to the original system
𝑲𝒛 = 𝑲𝒙 𝑷−1 = [−20 10 − 2]
• The characteristic equation 1
det 𝑠𝑰 − 𝑨𝒙 − 𝑩𝒙 𝑲𝒙 1 1/𝑠 1 1/𝑠 1 1/𝑠 4
𝑟 𝑦
−1 −2 −5
= 𝑠 3 + 8 + 𝑘3𝑥 𝑠 2 + 17 + 𝑘2𝑥 𝑠 + 10 + 𝑘1𝑥 𝑢
2
𝑧3 𝑧2 𝑧1

−10
=0 (12.52) 20
Designed system with state-variable feedback

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 45 Design via State Space System Dynamics and Control 46 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

Verify the design Run ch12p3 in Appendix B
The state equations for the designed system Learn how to use MATLAB to
−5 1 0 0 • design a controller for a plant not represented in
𝒛ሶ = 𝑨𝒛 − 𝑩𝒛𝑲𝒛 𝒛 + 𝑩𝒛𝑟 = 0 −2 1 𝒛 + 0 𝑟 phase-variable form
20 −10 1 1
𝑦 = 𝑪𝒛𝒛 = −1 1 0 𝒛 • see that MATLAB does not require transformation to
The closed-loop TF phase-variable form
𝑠+4 1 • solve Ex.12.4
𝑇 𝑠 = 3 =
𝑠 + 6𝑠 2 + 13𝑠 + 20 𝑠 2 + 2𝑠 + 5 1
1 1/𝑠 1 1/𝑠 1 1/𝑠 4
𝑟 𝑦
𝑢 −1 𝑧3 −2 𝑧2 −5 𝑧1
2
−10
20
Designed system with state-variable feedback

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 47 Design via State Space System Dynamics and Control 48 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

Skill-Assessment Ex.12.3 To cancel the zero at −6, adding a pole at −6 yields the
Problem Design a linear state-feedback controller to yield 20% resulting desired characteristic equation
overshoot and a settling time of 2𝑠 for a plant 𝑠 2 + 4𝑠 + 19.24 𝑠 + 6
𝑠+6 = 𝑠 3 + 10𝑠 2 + 43.24𝑠 + 115.45 = 0
𝐺 𝑠 =
(𝑠 + 9)(𝑠 + 8)(𝑠 + 7) 𝑠+6
Since 𝐺 𝑠 =
Solution First check controllability (𝑠 + 9)(𝑠 + 8)(𝑠 + 7)
0 0 1 𝑠+6
𝑪𝑴𝒛 = 𝑩 𝑨𝑩 𝑨𝟐 𝑩 = 0 1 −17 = 3
𝑠 + 24𝑠 2 + 191𝑠 + 504
1 −9 81
We can write the phase-variable representation
𝑪𝑴𝒛 = −1 ≠ 0 ⟹ rank 𝑪𝑴 = 3 : the system is
controllable 0 1 0 0
𝑨𝑝 = 0 0 1 ,𝑩𝑝 = 0 ,𝑪𝑝 = [6 1 0]
Now find the desired characteristic equation −504 −191 −24 1
𝑂𝑆 = 20% 𝜉 = 0.456
ቋ→
𝑇𝑠 = 2𝑠 𝜔𝑛 = 4.386
⟹ 𝑠 2 + 2𝜉𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 4𝑠 + 19.24 = 0 𝑠 2 + 4𝑠 + 19.24 = 0
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

8
 04/12/2020

System Dynamics and Control 49 Design via State Space System Dynamics and Control 50 Design via State Space

§4.Alternative Approaches to Controller Design §4.Alternative Approaches to Controller Design

The compensated system matrix in phase-variable form Now develop the transformation matrix to transform
0 1 0 back to the 𝑧-system
𝑨𝑝 − 𝑩𝑝𝑲𝑝 = 0 0 1 0 0 1
−(504 + 𝑘1) −(191 + 𝑘2) −(24 + 𝑘3) 𝑪𝑴𝒛 = 𝑩𝒛 𝑨𝒛 𝑩𝒛 𝑨2𝒛 𝑩𝒛 = 0 1 −17
1 −9 81
The characteristic equation for this system 0 0 1
𝑠𝑰 − 𝑨𝑝 − 𝑩𝑝 𝑲𝑝 2
𝑪𝑴𝒑 = 𝑩𝒑 𝑨𝒑𝑩𝒑 𝑨𝒑𝑩𝒑 = 0 1 −24
= 𝑠 3 + 24 + 𝑘3 𝑠 2 + 191 + 𝑘2 𝑠 + (504 + 𝑘1 ) Therefore 1 −24 385
1 0 0
Equating coefficients of this equation with the −1
𝑷 = 𝑪𝑴𝒛 𝑪𝑴𝒑 = 7 1 0
coefficients of the desired characteristic equation yields Hence, 56 15 1
the gains −1
𝑲𝒛 = 𝑲𝒑 𝑪𝑴𝒑
𝑲𝑝 = 𝑘1 𝑘2 𝑘3 1 0 0
= [−388.55 − 147.76 − 14] = [−388.55 − 147.76 − 14] 7 1 0
56 15 1
𝑠 3 + 10𝑠 2 + 43.24𝑠 + 115.45 = 0 = [−40.23 62.24 − 14]
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 51 Design via State Space System Dynamics and Control 52 Design via State Space

§5.Observer Design §5.Observer Design

- Controller design relies upon access to the state variables for - Let’s look at the disadvantages of such configuration. Assume the plant
feedback through adjustable gains 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢, 𝑦 = 𝑪𝒙 (12.57)
- Some of the state variables may not be available at all, or it is and an observer
too costly to measure them or send them to the controller
ෝሶ = 𝑨ෝ
𝒙 𝒙 + 𝑩𝑢, 𝑦ො = 𝑪ෝ
𝒙 (12.58)
- If the state variables are not available because of system
Subtracting Eqs. (12.58) from (12.57) to obtain
configuration or cost, it is possible to estimate the states.
Estimated states, rather than actual states, are then fed to the ෝሶ = 𝑨(𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ), 𝑦 − 𝑦ො = 𝑪(𝒙 − 𝒙
ෝ) (12.59)
controller. One scheme is shown in the figure Thus, the dynamics of the difference between the actual and
An observer, sometimes estimated states is unforced, and if the plant is stable, this
called an estimator, is used difference, due to differences in initial state vectors, approaches zero
to calculate state variables The speed of convergence 𝒙 → 𝒙 ෝ is too slow → using feedback
that are not accessible from to increase the speed of convergence between the actual and
the plant. Here the observer estimated states
is a model of the plant

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 53 Design via State Space System Dynamics and Control 54 Design via State Space

§5.Observer Design §5.Observer Design

- Use feedback to increase the speed of convergence between - In designing a controller, the controller canonical (phase-
the actual and estimated states variable) form yields an easy solution for the controller gains
plant output The error between the outputs of In designing an observer, the observer canonical form yields
plant 𝑦
𝑟=0 𝑢 the plant and the observer is fed the easy solution for the observer gains
observer back to the derivatives of the
controller
estimated states
ෝ
𝒙 observer’s states. The system
corrects to drive this error to zero
With feedback we can design a desired transient response into
estimated output the observer that is
𝑢 ෝሶ
𝒙 ෝ
𝒙 ෝ
𝒚 𝑦
𝑩 𝑪 plant output much quicker than

𝑨 to controller
that of the plant or
controlled closed-
𝑳
estimated error output loop system

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

9
 04/12/2020

System Dynamics and Control 55 Design via State Space System Dynamics and Control 56 Design via State Space

§5.Observer Design §5.Observer Design

- Example a third-order plant - The design of the observer is separate from the design of the
• represented in observer canonical form controller
𝑏3
𝑏2 - Similar to the design of the controller vector, 𝑲, the design of
𝑏1 1/𝑠 1 1/𝑠 1 1/𝑠 1
the observer consists of evaluating the constant vector, 𝑳, so
𝑢 𝑦ො
𝑥ො3 𝑥ො2 −𝑎2 𝑥ො1 that the transient response of the observer is faster than the
−𝑎1
−𝑎0 response of the controlled loop in order to yield a rapidly
Third-order observer in observer canonical form before the addition of the feedback updated estimate of the state vector
• configured as an observer with the addition of feedback • Find the state equations for the error between the actual state
𝑏3 ෝ
vector and the estimated state vector, 𝒙 − 𝒙
𝑏2

𝑏1 1/𝑠 1 1/𝑠 1 1/𝑠 1 −1 1

• Find the characteristic equation for the error system and
𝑢 𝑦
𝑥ො3 𝑥ො2 −𝑎2 𝑥ො1 𝑦ො evaluate the required 𝑳 to meet a rapid transient response for
−𝑎1 𝑙1
−𝑎0 𝑙2 the observer
𝑙3
Third-order observer in observer canonical form after the addition of the feedback

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 57 Design via State Space System Dynamics and Control 58 Design via State Space

§5.Observer Design §5.Observer Design

- The state equations of the observer ෝሶ = (𝑨 − 𝑳𝑪)(𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ), 𝑦 − 𝑦ො = 𝑪(𝒙 − 𝒙
ෝ) (12.63)
ෝሶ = 𝑨ෝ
𝒙 𝒙 + 𝑩𝑢 + 𝑳 𝑦 − 𝑦ො , 𝑦ො = 𝑪ෝ
𝒙 (12.60) or 𝒆ሶ 𝒙 = (𝑨 − 𝑳𝑪)𝒆𝒙 , 𝑦 − 𝑦ො = 𝑪(𝒙 − 𝒙
ෝ) (12.64)
- The state equations for the plant 𝒆𝒙 : the estimated state error, 𝒆𝒙 = 𝒙 − 𝒙 ෝ
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢, 𝑦 = 𝑪𝒙 (12.61) - Eq. (12.64a) is unforced. If the eigenvalues are all negative, the
- Subtracting Eqs. (12.60) from (12.61) to obtain estimated state vector error, 𝒆𝒙 , will decay to zero. The design
then consists of solving for the values of 𝑳 to yield a desired
ෝሶ = 𝑨 𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ − 𝑳(𝑦 − 𝑦), ෝ)
ො 𝑦 − 𝑦ො = 𝑪(𝒙 − 𝒙 (12.62)
characteristic equation or response for Eq. (12.64). The
𝒙−𝒙 ෝ : the error between the actual state vector and the characteristic equation is found from Eq. (12.64) to be
estimated state vector
det 𝜆𝑰 − 𝑨 − 𝑳𝑪 =0 (12.65)
𝑦 − 𝑦ො : the error between the actual output and the estimated
output Now we select the eigenvalues of the observer to yield stability
and a desired transient response that is faster than the
- Subtracting the output equation into the state equation to obtain
controlled closed-loop response. These eigenvalues determine
ෝሶ = (𝑨 − 𝑳𝑪)(𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ), 𝑦 − 𝑦ො = 𝑪(𝒙 − 𝒙
ෝ) (12.63) a characteristic equation that we set equal to Eq. (12.65) to
solve for 𝑳
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 59 Design via State Space System Dynamics and Control 60 Design via State Space

§5.Observer Design §5.Observer Design

- Ex.12.5 Observer Design for Observer Canonical Form 1.First represent the estimated plant in observer canonical form
Design an observer for the plant 1

𝑠+4 𝑠 +4 4 1/𝑠 1 1/𝑠 1 1/𝑠 1

𝐺 𝑠 = = 𝑢
𝑥ො3 𝑥ො2 −8 𝑥ො1
𝑦ො
(𝑠 + 1)(𝑠 + 2)(𝑠 + 5) 𝑠 3 + 8𝑠 2 + 17𝑠 + 10 −17
−10
which is represented in observer canonical form. The observer
will respond 10 times faster than the controlled loop designed in 2.Now form the difference between the plant’s actual output, 𝑦,
Ex.12.4 and the observer’s estimated output, 𝑦,
ො and add the feedback
Solution paths from this difference to the derivative of each state
variable
1.First represent the estimated plant in observer canonical form
1
1
4 1/𝑠 1 1/𝑠 1 1/𝑠 1 −1 1
4 1/𝑠 1 1/𝑠 1 1/𝑠 1 𝑢 𝑦
𝑢 𝑦ො 𝑥ො3 𝑥ො2 −8 𝑥ො1 𝑦ො
𝑥ො3 𝑥ො2 −8 𝑥ො1 −17 𝑙1
−17 −10 𝑙2
−10 𝑙3

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

10
 04/12/2020

System Dynamics and Control 61 Design via State Space System Dynamics and Control 62 Design via State Space

§5.Observer Design §5.Observer Design

3.Next find the characteristic polynomial. The state equations for the 4.Now evaluate the desired polynomial, set the coefficients
estimated plant 1 equal to those of Eq. (12.74), and solve for the gains, 𝑙𝑖 . From
4 1/𝑠 1 1/𝑠 1 1/𝑠 1
Eq. (12.50), the closed-loop controlled system has dominant
𝑢 𝑦ො
𝑥ො3 𝑥ො2 −8 𝑥ො1 second-order poles at −1 ± 𝑗2. To make our observer 10
−17
−10 times faster, we design the observer poles to be at −10 ±
𝑗20. We select the third pole to be 10 times the real part of
−8 1 0 0 the dominant second-order poles, or −100 . Hence, the
ොሶ = 𝑨ො𝒙 + 𝑩𝑢 = −17 0 1 𝒙
𝒙 ො + 1 𝑢, 𝑦Ƹ = 𝑪ො𝒙 = 1 0 0 𝒙
ො desired characteristic polynomial
−10 0 0 4
𝑠 + 100 𝑠 2 + 20𝑠 + 500 =
The observer error
−(8 + 𝑙1 ) 1 0 𝑠 3 + 120𝑠 2 + 2500𝑠 + 50,000 = 0 (12.75)
𝒆ሶ 𝑥 = 𝑨 − 𝑳𝑪 𝒆𝑥 = −(17 + 𝑙2 ) 0 1 𝒆𝑥 Equating Eqs. (12.74) and (12.75) to obtain
−(10 + 𝑙3 ) 0 0 𝑙1 = 112, 𝑙2 = 2483, 𝑙3 = 49,990
The characteristic polynomial
𝑠 + 4 𝑠 2 + 2𝑠 + 5 = 𝑠 3 + 6𝑠 2 + 13𝑠 + 20 = 0 (12.50)
𝑠 3 + 8 + 𝑙1 𝑠 2 + 17 + 𝑙2 𝑠 + 10 + 𝑙3 = 0 (12.74) 𝑠 3 + 8 + 𝑙1 𝑠 2 + 17 + 𝑙2 𝑠 + 10 + 𝑙3 = 0 (12.74)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 63 Design via State Space System Dynamics and Control 64 Design via State Space

§5.Observer Design §5.Observer Design

A simulation of the observer with an input of 𝑟 𝑡 = 100𝑡 is Run ch12p4 in Appendix B
shown in the figure. The initial conditions of the plant were all Learn how to use MATLAB to
zero, and the initial condition of 𝑥ො1 was 0.5 • design an observer using pole placement
• solve Ex.12.5

Since the dominant pole of the observer is −10 ± 𝑗20, the

expected settling time should be about 0.4𝑠. It is interesting to
note the slower response in the figure, where the observer
gains are disconnected, and the observer is simply a copy of
the plant with a different initial condition
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 65 Design via State Space System Dynamics and Control 66 Design via State Space

§5.Observer Design §5.Observer Design

Skill-Assessment Ex.12.4 Solution The plant is given by
Problem Design an observer for the plant 𝑠+6 20
𝐺 𝑠 = =
𝑠+6 (𝑠 + 9)(𝑠 + 8)(𝑠 + 7) 𝑠3 + 14𝑠2 + 56𝑠 + 64
𝐺 𝑠 =
(𝑠 + 9)(𝑠 + 8)(𝑠 + 7) The characteristic polynomial for the plant with phase-
whose estimated plant is represented in state space in variable state feedback
observer canonical form as 𝑠 3 + (𝑘3 + 14)𝑠 2 + (𝑘2 + 56)𝑠 + (𝑘3 + 64) = 0
−24 1 0 0 The desired characteristic equation
ෝሶ = 𝑨ෝ
𝒙 𝒙 + 𝑩𝑢 = −191 0 1 𝒙 ෝ+ 1 𝑢 𝑠 + 53.33 𝑠2 + 10.67𝑠 + 106.45 =
−504 0 0 6
𝑠 3 + 64𝑠 2 + 675.48𝑠 + 5676.98 = 0
𝑦ො = 𝑪ෝ𝒙= 1 0 0 𝒙 ෝ
The observer will respond 10 times faster than the based upon 15% overshoot, 𝑇𝑠 = 0.75𝑠, and a third
controlled loop designed in Skill-Assessment Ex.12.3 pole ten times further from the imaginary axis than the
dominant poles
Comparing the two characteristic equations
𝑘1 = 5612.98, 𝑘2 = 619.48, and 𝑘3 = 50
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

11
 04/12/2020

System Dynamics and Control 67 Design via State Space System Dynamics and Control 68 Design via State Space

§5.Observer Design §6.Observability

TryIt 12.3 −24 1 0 0 Recall that the ability to control all of the state variables is a
Use MATLAB, the Control ෝሶ = 𝑨ෝ
𝒙 ෝ+ 1 𝑢
𝒙 + 𝑩𝑢 = −191 0 1 𝒙 requirement for the design of a controller. State-variable feedback
System Toolbox, and the −504 0 0 6 gains cannot be designed if any state variable is uncontrollable.
following statements to solve
𝒙= 1 0 0 ෝ
𝑦ො = 𝑪ෝ 𝒙 Uncontrollability can be viewed best with diagonalized systems.
Skill-Assessment Ex.12.4
The signal-flow graph showed clearly that the uncontrollable state
A=[-24 1 0; -191 0 1; -504 0 0];
C=[1 0 0]
variable was not connected to the control signal of the system
pos=20 • Controllable system 1/𝑠
1 𝑘1
Ts=2 −𝑎1 𝑥1
z=(-log(pos/100))/(sqrt(pi^2 +log(pos/100)^2)); −𝑎1 0 0 1
𝑢 1 1/𝑠 𝑘2 𝑦
wn=4/(z*Ts); 𝒙ሶ = 0 −𝑎2 0 𝒙+ 1 𝑢 −𝑎2 𝑥2
r=roots([1,2*z*wn,wn^2]); 0 0 −𝑎3 1
1/𝑠
poles=10*[r' 10*real(r(1))] 1 −𝑎3 𝑘3
𝑥3
l=acker(A',C',poles)' 1/𝑠
• Uncontrollable system 𝑘4
−𝑎4 𝑥1
−𝑎4 0 0 0
𝑢 1 1/𝑠 𝑘5 𝑦
𝒙ሶ = 0 −𝑎5 0 𝒙+ 1 𝑢 −𝑎5 𝑥2
0 0 −𝑎6 1
1/𝑠
1 −𝑎6 𝑘6
𝑥3
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 69 Design via State Space System Dynamics and Control 70 Design via State Space

§6.Observability §6.Observability
1/𝑠 1/𝑠 1/𝑠 1/𝑠
𝑘1 1 𝑘4 𝑘1 1 𝑘4
−𝑎1 𝑥1 −𝑎4 𝑥1 −𝑎1 𝑥1 −𝑎4 𝑥1

𝑢 𝑘2 1/𝑠 1 𝑦 𝑢 𝑘5 1/𝑠 1 𝑦 𝑢 𝑘2 1/𝑠 1 𝑦 𝑢 𝑘5 1/𝑠 1 𝑦

−𝑎2 𝑥2 −𝑎5 𝑥2 −𝑎2 𝑥2 −𝑎5 𝑥2
1/𝑠 1/𝑠 1/𝑠 1/𝑠
𝑘3 −𝑎3 1 𝑘6 −𝑎6 1 𝑘3 −𝑎3 1 𝑘6 −𝑎6 1
𝑥3 𝑥3 𝑥3 𝑥3
observable system unobservable system observable system unobservable system

The ability to observe a state variable from the output is best Simply stated, observability is the ability to deduce the state
seen from the diagonalized system. Here 𝑥1 is not connected to variables from a knowledge of the input, 𝑢(𝑡), and the output,
the output and could not be estimated from a measurement of 𝑦(𝑡)
the output Pole placement for an observer is a viable design technique
If the initial-state vector, 𝑥(𝑡0 ), can be found from 𝑢(𝑡) and only for systems that are observable
𝑦(𝑡) measured over a finite interval of time from 𝑡0 , the system
is said to be observable; otherwise the system is said to be
unobservable

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 71 Design via State Space System Dynamics and Control 72 Design via State Space

§6.Observability §6.Observability
𝑘1
1/𝑠
1 𝑘4
1/𝑠 The Observability Matrix
−𝑎1 𝑥1 −𝑎4 𝑥1

𝑢 𝑘2 1/𝑠 1 𝑦 𝑢 𝑘5 1/𝑠 1 𝑦
An 𝑛th-order plant whose state equation is
−𝑎2 𝑥2 −𝑎5 𝑥2
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
1/𝑠 1/𝑠
𝑘3 −𝑎3 1 𝑘6 −𝑎6 1 is completely observable if the matrix
𝑥3 𝑥3
observable system unobservable system 𝑶𝑴 = [𝑪 𝑪𝑨 𝑪𝑨𝟐 ⋯ 𝑪𝑨𝒏−𝟏 ]𝑇
Observability by Inspection is of rank 𝑛, where 𝑶𝑴 is called the observability matrix
The system can be explored from the output equation of a
diagonalized system
• for the observable system
𝑦 = 𝑪𝒙 = 1 1 1 𝒙
• for the unobservable system
𝑦 = 𝑪𝒙 = 0 1 1 𝒙

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

12
 04/12/2020

System Dynamics and Control 73 Design via State Space System Dynamics and Control 74 Design via State Space

§6.Observability §6.Observability
- Ex.12.6 Observability via the Observability Matrix Run ch12p5 in Appendix B
1
Determine if the system is observable 5 Learn how to use MATLAB to
𝑢
1 1/𝑠 1/𝑠 1/𝑠
𝑦
• test a system for observability
−2 𝑥3 𝑥2 𝑥1
−3 • solve Ex.12.6
Solution −4

The state and output equations for the system

0 1 0 0
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 0 0 1 𝒙+ 0 𝑢
−4 −3 −2 1
𝑦 = 𝑪𝒙 = 0 5 1 𝒙
The observability matrix
𝑪 0 5 1
𝑶𝑴 = 𝑪𝑨 = −4 −3 3
𝑪𝑨𝟐 −12 −13 −9
det 𝑶𝑴 = −344, rank 𝑶𝑴 = 3 ⟹ system is observable
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 75 Design via State Space System Dynamics and Control 76 Design via State Space

§6.Observability §6.Observability
- Ex.12.7 Unobservability via the Observability Matrix Skill-Assessment Ex.12.5
Determine if the system is observable Problem Determine whether the system
4
−2 1 −3 2
𝑢
1 1/𝑠 1/𝑠 5
𝑦
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 0 −2 1 𝒙 + 1 𝑢
−21/4 𝑥2 𝑥1
−5 −7 −8 −9 2
Solution 𝑦 = 𝑪𝒙 = 4 6 8 𝒙
The state and output equations for the system is observable
0 1 0
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝒙+ 𝑢 Solution The observability matrix
−5 −21/4 1 𝑪 4 6 8
𝑦 = 𝑪𝒙 = 5 4 𝒙 𝑶𝑴 = 𝑪𝑨 = −64 −80 −78
The observability matrix 𝟐
𝑪𝑨 674 848 814
𝑪 5 4 det 𝑶𝑴 = −1576, rank 𝑶𝑴 = 3
𝑶𝑴 = =
𝑪𝑨 −20 −16 ⟹ system is observable
det 𝑶𝑴 = 0, rank 𝑶𝑴 < 3 ⟹ system is unobservable

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 77 Design via State Space System Dynamics and Control 78 Design via State Space

§6.Observability §7.Alternative Approaches to Observer Design

TryIt 12.4 −2 1 −3 2 - Assume a plant not represented in observer canonical form
Use MATLAB, the Control 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 =
0 −2 1 𝒙 + 1 𝑢
𝒛ሶ = 𝑨𝒛 + 𝑩𝑢, 𝑦 = 𝑪𝒛 (12.84)
System Toolbox, and the −7 −8 −9 2
following statements to solve
𝑦 = 𝑪𝒙 = 4 6 8 𝒙 - The observability matrix
Skill-Assessment Ex.12.5
𝑶𝑴𝒛 = [𝑪 𝑪𝑨 𝑪𝑨𝟐 ⋯ 𝑪𝑨𝒏−𝟏 ]𝑇 (12.85)
A = [-2 -1 -3; 0 -2 1; -7 -8 -9] - Now assume that the system can be transformed to the
C=[4 6 8]
Om=obsv(A,C) observer canonical form, 𝒙, with the transformation
Rank=rank(Om) 𝒛 = 𝑷𝒙 (12.86)
- Substituting Eq. (12.86) into Eqs. (12.84) and pre-multiplying
the state equation by 𝑷−1 , we find that the state equations in
observer canonical form are

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

13
 04/12/2020

System Dynamics and Control 79 Design via State Space System Dynamics and Control 80 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
- Consider the controlled system - Rewritten as augmented vectors and matrices
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢, 𝑦 = 𝑪𝒙 (12.112) 𝒙ሶ 𝑨 𝟎 𝒙 𝑩 𝟎 𝒙
= + 𝑢+ 𝑟, 𝑦 = 𝑪 0 𝑥 (12.113)
𝑟 𝑢
𝑩
𝒙ሶ 𝒙
𝑪
𝑦 𝑥𝑁ሶ −𝑪 𝟎 𝑥𝑁 0𝑁 1 𝑁
but
𝑨 𝒙
𝑢 = −𝑲𝒙 + 𝐾𝑒 𝑥𝑁 = − 𝑲 −𝐾𝑒 𝑥 (12.114)
𝑲 𝑁

- An additional state variable 𝑥𝑁 has been added at the output of - Substituting Eq. (12.114) into (12.113) and simplifying
the leftmost integrator. The error is the derivative of this variable 𝒙ሶ 𝑨 − 𝑩𝑲 𝑩𝐾𝑒 𝒙 𝟎 𝒙
= + 𝑟, 𝑦 = 𝑪 0 𝑥 (12.115)
𝑥𝑁 ሶ −𝑪 𝟎 𝑥𝑁 1 𝑁
𝑥ሶ 𝑁 = 𝑟 − 𝑪𝒙 (12.111)
𝑟 𝑒 𝑥𝑁 𝑢 𝒙ሶ 𝒙 𝑦 The system type has been increased, and we can use the
𝑲𝒆 𝑩 𝑪 characteristic equation associated with Eq. (12.115) to design
𝑨
𝑲 and 𝐾𝑒 to yield the desired transient response

𝑲
𝑥ሶ 𝑁 = 𝑟 − 𝑪𝒙 = −𝑪𝒙 + 𝒓 (12.111), 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢, 𝑦 = 𝑪𝒙 (12.112)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 81 Design via State Space System Dynamics and Control 82 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
- Ex.12.10 Design of Integral Control 𝑠 2 + 16𝑠 + 183.1 (12.117)
Consider the plant 𝑠 2 + (5 + 𝑘2 )𝑠 + (3 + 𝑘1) (12.118)
0 1 0 Equating the coefficients of the above two Eqs. to get
𝒙ሶ = 𝒙+ 𝑢, 𝑦 = 1 0 𝒙 (12.16)
−3 −5 1
𝐾 = 𝑘1 𝑘2 = 180.1 11 (12.19)
a.Design a controller without integral control to yield a 10%
From Eqs. (12.3), the controlled plant with state-variable
overshoot and a settling time of 0.5𝑠. Evaluate the steady-
feedback represented in phase variable form
state error for a unit step input
0 1 0
b. Repeat the design of (a) using integral control. Evaluate the 𝒙ሶ = 𝑨 − 𝑩𝑲 𝒙 + 𝑩𝑟 = 𝒙+ 𝑟 (12.120)
−183.1 −16 1
steady-state error for a unit step input 𝑦 = 𝑪𝒙 = 1 0 𝒙
Solution Using Eq.(7.96), the steady-state error for a step input
a.From 𝑇𝑠 and %𝑂𝑆 the desired characteristic polynomial 0 1
−1
0
𝑒 ∞ =1+ 1 0 = 0.995 (12.121)
𝑠 2 + 16𝑠 + 183.1 (12.117) −183.1 −16 1
The characteristic polynomial for the controlled plant 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝑨𝒙 + 𝑩 −𝑲𝒙 + 𝑟 = 𝑨 − 𝑩𝑲 𝑥 + 𝑩𝑟, 𝑦 = 𝑪𝒙 (12.3)
𝑠 2 + (5 + 𝑘2 )𝑠 + (3 + 𝑘1) (12.118) 𝑒 ∞ = 1 − 𝑦𝑠𝑠 = 1 − 𝑪𝑽 = 1 + 𝑪𝑨−1 𝑩 (7.96)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 83 Design via State Space System Dynamics and Control 84 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
b.Use Eqs. (12.115) to represent the integral-controlled plant 𝑥ሶ 1 0 1 0 𝑥1 0
𝑥ሶ 1 0 1 0 0 𝑥1 0 𝑥ሶ 2 = −(3 + 𝑘1 ) −(5 + 𝑘2 ) 𝐾𝑒 𝑥2 + 0 𝑟
− [𝑘 𝑘2 ] 𝐾 𝑥ሶ 𝑁 0 𝑥𝑁
𝑥ሶ 2 = −3 −5 1 1 1 𝑒 𝑥2 + 0 𝑟 −1 0 1
𝑥ሶ 𝑁 −[1 0] 0 𝑥𝑁 1 𝑥1
0 1 0 𝑥1 0 𝑦 = 1 0 0 𝑥2
= −(3 + 𝑘1 ) −(5 + 𝑘2 ) 𝐾𝑒 𝑥2 + 0 𝑟 𝑥𝑁
−1 0 0 𝑥𝑁 1 Augment (12.117) with a third pole, 𝑠 + 100, which has a real part
𝑥1 greater than five times that of the desired dominant 2nd-order poles.
𝑦 = 1 0 0 𝑥2 (12.122) The desired 3rd-order closed-loop system characteristic polynomial
𝑥𝑁 𝑠 + 100 𝑠 2 + 16𝑠 + 183.1
1
Using (12.16), (3.73), the TF of the plant, 𝐺 𝑠 = 2 = 𝑠 3 + 116𝑠 2 + 1783.1𝑠 + 18,310 (12.123)
𝑠 + 5𝑠 + 3
The characteristic polynomial for the system of Eqs. (12.112)
0 1 0
𝒙ሶ =
−3 −5
𝒙+
1
𝑢, 𝑦 = 1 0𝒙 (12.16) 𝑠 3 + (5 + 𝑘2 )𝑠 2 + (3 + 𝑘1 )𝑠 + 𝐾𝑒 (12.124)
𝑌(𝑠) −1 𝑩 + 1
𝑇 𝑠 = 𝑈(𝑠)
= 𝑪 𝑠𝑰 − 𝑨 𝑫 (3.73) 𝐺 𝑠 = 𝑠2 +5𝑠+3 , 𝑠 2 + 16𝑠 + 183.1 = [𝑠 − (8 − 10.9133𝑖)][𝑠 − (8 + 10.9133𝑖)] (12.117)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

14
 04/12/2020

System Dynamics and Control 85 Design via State Space System Dynamics and Control 86 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
𝑠 + 100 𝑠 2 + 16𝑠 + 183.1 In order to check our assumption for the zero, now find the
= 𝑠 3 + 116𝑠 2 + 1783.1𝑠 + 18,310 (12.123) closed-loop TF
18,310
𝑠 3 + (5 + 𝑘2 )𝑠 2 + (3 + 𝑘1 )𝑠 + 𝐾𝑒 (12.124) 𝑇 𝑠 = 𝑪 𝑠𝑰 − 𝑨 −1𝑩 + 𝑫 = 3 (12.127)
𝑠 +116𝑠2 +1783.1𝑠+18,310
Matching coefficients from Eqs. (12.123) and (12.124)
The TF matches our design ⟹ The desired transient response
𝑘1 = 1780.1, 𝑘2 = 111, 𝐾𝑒 = 18,310 (12.125)
The steady-state error for a unit step input
Substituting these values into Eqs. (1.122) yields this closed-
loop integral control system 𝑒 ∞ = 1 + 𝑪𝑨−1 𝑩 (7.96)
−1
𝑥ሶ 1 𝑥1 0 1 0 0
0 1 0 0
𝑥ሶ 2 = −1783.1 −116 18,310 𝑥2 + 0 𝑟 = 1 + 1 0 0 −1783.1 −116 18,310 0 =0
𝑥ሶ 𝑁 𝑥𝑁 −1 0 0 1
−1 0 0 1
The system behaves like Type 1 system (12.128)
𝑦 = 1 0 0 𝑥1 𝑥2 𝑥3 𝑇 (12.122)
𝑥ሶ 1 0 1 0 𝑥1 0 𝑥1 𝑥ሶ 1 0 1 0 𝑥1 0 𝑥1
𝑥ሶ 2 = −(3 + 𝑘1 ) −(5 + 𝑘2 ) 𝐾𝑒 𝑥2 + 0 𝑟, 𝑦 = 1 0 0 𝑥2 (12.122) 𝑥ሶ 2 = −(3 + 𝑘1 ) −(5 + 𝑘2 ) 𝐾𝑒 𝑥2 + 0 𝑟, 𝑦 = 1 0 0 𝑥2 (12.122)
𝑥ሶ 𝑁 −1 0 0 𝑥𝑁 1 𝑥𝑁 𝑥ሶ 𝑁 −1 0 0 𝑥𝑁 1 𝑥𝑁
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 87 Design via State Space System Dynamics and Control 88 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
Skill-Assessment Ex.12.7 𝑠 2 + 2.3𝑠 + 3.79 = 0 (*)
Problem Design an integral controller for the plant The characteristic polynomial for the controlled plant
0 1 0 𝑠 2 + 9 + 𝑘2 𝑠 + 7 + 𝑘1 = 0 (**)
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝒙+ 𝑢
−7 −9 1 Equating the coefficients of the above two Eqs. to get
𝑦 = 𝑪𝒙 = 4 1 𝒙
𝐾 = 𝑘1 𝑘2 = −3.21 −6.70
to yield a step response with 10% overshoot, a peak
The controlled plant with state-variable feedback
time of 2s and zero steady-state error
0 1 0
Solution The desired characteristic equation 𝒙ሶ = 𝑨 − 𝑩𝑲 𝒙 + 𝑩𝑟 = 𝒙+ 𝑟
−3.79 −2.30 1
ln %𝑂𝑆/100 ln 10/100 𝑦 = 𝑪𝒙 = 4 1 𝒙
𝜉=− =− = 0.591
𝜋2 + 𝑙𝑛2 %𝑂𝑆/100 𝜋2 + 𝑙𝑛2 10/100 The steady-state error for a step input
𝜋 4 0 1
−1
0
𝜔𝑛 = = = 1. 948𝑟𝑎𝑑/𝑠 𝑒 ∞ =1+ 4 1 = −0.0554
𝑇𝑝 1 − 𝜉 2 2 1 − 0. 5912 −3.79 −2.30 1
⟹ 𝑠 2 + 2𝜉𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 2.3𝑠 + 3.79 = 0 (*) 0 1 0
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝒙+ 𝑢, 𝑦 = 𝑪𝒙 = 4 1 𝒙, 𝑒 ∞ = 1 − 𝑦𝑠𝑠 = 1 + 𝑪𝑨−1 𝑩
−7 −9 1
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

System Dynamics and Control 89 Design via State Space System Dynamics and Control 90 Design via State Space

§8.Steady-State Error Design via Integral Control §8.Steady-State Error Design via Integral Control
The integral-controlled plant The TF of the original system
𝑥ሶ 1 0 1 0 𝑥1 0 𝐺 𝑠 = 𝑪 𝑠𝑰 − 𝑨 −1 𝑩
𝐾
𝑥ሶ 2 = −3.79 −2.30 1 𝑒 𝑥2 + 0 𝑟 𝑠 −1
−1
0
𝑥ሶ 𝑁 −[4 1] 0 𝑥𝑁 = 4 1
1 7 𝑠+9 1
0 1 0 𝑥1 0 𝑠+9 1
= −3.79 −2.30 𝐾𝑒 𝑥2 + 0 𝑟 = 4 1
−7 𝑠 0
−4 −1 0 𝑥𝑁 1 𝑠 𝑠+9 +7 1
𝑥1 𝑠+4
𝑦 = 1 0 0 𝑥2 = 2
𝑠 + 9𝑠 + 7
𝑥𝑁
Adding a pole at −4 which corresponds to the original
system’s zero location, yield the characteristic equation
𝑠2 + 2.3𝑠 + 3.97 = 0 𝑠 + 4 = 𝑠3 + 6.3𝑠2 + 13𝑠 + 15.16

𝒙ሶ 𝑨 − 𝑩𝑲 𝑩𝐾𝑒 𝒙 𝟎 𝒙 0 1 0
= + 𝑟, 𝑦 = 𝑪 0 𝑥 (12.115) 𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 = 𝒙+ 𝑢, 𝑦 = 𝑪𝒙 = 4 1 𝒙
𝑥𝑁ሶ −𝑪 𝟎 𝑥𝑁 1 𝑁 −7 −9 1
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

15