Basic Elements of a Control System Advantages of closed-loop system

- output is less sensitive to disturbances and

plant changes
- Disturbances
o external factors
- Plant Changes
o internal changes

Disadvantages of closed-loop system

1. Plant
- more expensive
- process you want to control
- must be familiar with physics of process - unstable if improperly designed
- requires knowledge of: Control System Design Flowchart
a. basic energy balance 1. control objectives
b. mass balance
- experience, dialogue
c. material flows
2. modeling
- system identification (science, math)
2. Sensor
3. model verification
- “eye” of the control
- reports on the state of the process - simulation
4. control architecture
3. Actuator - experience
- moves process from current state to the 5. control design
desired state. - root locus, bode, Nyquist, simulation
- “muscle” 6. control verification
- simulation, prototyping
4. Controller
- decides how to actuate the plant to achieve What do you want control to achieve?
desired state. - enhance quality
- energy reduction
Improving control yields:
- increase yield
- enhanced product quality
- less waste What level of performance is necessary?
- leads to environment friendliness - accuracy (error < 5%)
- greater throughput for a given installed - space (e.g. system must respond in < 1 min.)
capacity
- greater yield Model
- deferring costly plant upgrades - usually expressed in mathematical form
- higher safety margins - must be simple enough but enough to
provide good understanding of how process
Open-Loop System vs Close-Loop System works

Open-Loop System What should be considered?

- no feedback - classical, fuzzy, neutral
- uses “inversion” - does it consider modeling errors?
- how does it react to disturbances and noise?
Close-Loop System
- cost and complexity
- with feedback
Basic Math Tools State-Space Representation

Dynamic Systems are usually represented or First-Order ODE

modeled by differential equations.
1. Linear ODE
2. non-linear ODE
3. partial differential equations (PDE)

How to solve ODE?

1. Classical approach y’ = Ay + Bf
2. Laplace Transform where:
f = input
N-th order differential equation (DE)
Second-Order ODE

a3y’’ + a2y’ + a1y = f

- homogenous if f(t) = 0
−1 1
𝑦 ′′ = (𝑎2 𝑦′ + 𝑎1 𝑦) + ( ) 𝑓
𝑎3 𝑎3
Non-Linear differential equations
Defining vector:
𝑦 𝑦′
𝑥 = [𝑦′] 𝑥′ = [ ′ ]
𝑦′
Substituting:
𝑦′ 𝑦′ 0
′
𝑥 = [ ′ ] = [−1 ′ ] + [( 1 ) 𝑓 ]
𝑦′ (𝑎2 𝑦 + 𝑎1 𝑦) 𝑎3
𝑎3

1 0 0
𝑥 = [− 𝑎2
′
−
𝑎1 ] 𝑥 + [ 1 ]
( )𝑓
𝑎3 𝑎3 𝑎3

Linear Time-Invariant Systems Linear Approximations

- non-linear may be linear about an operating
Linear Systems point
- satisfies the following conditions:
o superposition Mechanical and Electrical elements
- linear over large range of variables

Thermal and Fluid elements

o homogeneity - highly nonlinear

Taylor Series Expansion (review on onenote)

Time-Invariant Systems
- time-shift independent
Model 4. Spring (linear)
- process of representing real-world object or f = kx
phenomenon as a set of mathematical
equations.
Types:
1. Verbal Model
2. Graphical Model example: Mass, spring, and damper system
3. Mathematical Model
- used in 151 f= Mx’’ + kx + Bx’
- may start from verbal or graphical to extract
an approximate mathematical model

Dynamic Systems
- systems where variables change with
respect to time
- usually modeled with differential equations

Electrical Systems 5. Rotational Variables

- Resistance, Capacitance, Inductance a. angle, θ
b. angular velocity, θ’
c. angular acceleration, θ’’
d. torque, T

6. Inertia (Rotational)
T = Jθ’’

Mechanical Systems
1. Translational Variable
a. position, x
b. velocity, x’ where:
c. acceleration, x’’ J = ½ Mr2 (for circular
d. force, f disk rotating about a geometric
2. Mass axis)
f = Mx’’ 7. Friction (rotational)
T = Bθ’

3. Friction (viscous)
f = Bx’
8. Spring (torsional)
T=k
 9. Gears Heat increase from entering liquid
𝑻𝟏 𝜽𝟐 𝑵𝟏 𝜽′𝟐
= = =
𝑻𝟐 𝜽𝟏 𝑵𝟐 𝜽′𝟏
where:
v(t) = liquid flow rate in and out
H = specific heat capacity

Heat Loss Through Tank

10. Belt-Driven Gears where:

𝑻𝟏 𝜽𝟐 𝒓𝟏 𝜽′𝟐 R = thermal resistance
= = =
𝑻𝟐 𝜽𝟏 𝒓𝟐 𝜽′𝟏
Differential equation of model

Assuming flow rate v(t) is constant value

V, we get a time invariant first-order ODE
REVIEW EXAMPLE ON SLIDES AND ONENOTE

Thermal Systems and Liquid-Level Systems

where:
qe(t) → input
τi(t) and τa(t) → disturbances
τ(t) → output

Some definitions
1. Nominal model
- approximate description
- used in doing control
2. Calibration model
- comprehensive description
- used to verify if control works
3. Model error
- difference between nominal and calibration
Conservation of energy model.

Heat Absorbed

where: C = thermal capacity

dτ(t)/dt = rate of temp. change
Block Diagram Transformations Forcing Functions
(get from the online link!!!) 1. Step input
2. Ramp input
Closed-Loop Transfer Function 3. Parabolic input
4. Sinusoids

OCTAVE!!! (set up file??)

Signal Flow Graphs

from error equation and output equation: - line segment rather than blocks
𝒀(𝒔) 𝑮(𝒔)
=
𝑹(𝒔) 𝟏 + 𝑮(𝒔)𝑯(𝒔) Signal Flow Graph Transformations
GET FROM LINK
Laplace Transform (get from book!!!)
Mason Gain Rule
Initial Value Theorem OneNote!!

Transfer Functions

Final Value Theorem

𝒀(𝒔) = 𝑮(𝒔)𝑹(𝒔)
𝑵(𝒔) [𝒏𝒖𝒎𝒆𝒓𝒂𝒕𝒐𝒓]
𝑮(𝒔) =
Laplace Transform Table 𝑫(𝒔)[𝒅𝒆𝒏𝒐𝒎𝒊𝒏𝒂𝒕𝒐𝒓]
Zeroes
- roots of N(s)
Poles
- roots if D(s)

Open-Loop System
- poles affect system response
𝑵(𝒔)
𝑮(𝒔) =
(𝒔 − 𝒑𝟏 ) … (𝒔 − 𝒑𝒏 )
impulse response:
𝒑 𝒕
𝒚(𝒕) = 𝒌𝟏𝟏 + ⋯ + 𝒌𝒏 𝒆𝒑𝒏 𝒕

Closed-Loop System
- poles and zeros affect the system response

𝒀(𝒔) 𝑮(𝒔) 𝑵(𝒔)

= =
𝑹(𝒔) 𝟏 + 𝒌𝑮(𝒔) 𝑫(𝒔) + 𝒌𝑵(𝒔)

General Control System

PARTIAL FRACTION EXPANSION
 - small U
- needs small KS, approx. 1/G for large K

Typical Control System Design

1. Large loop gain at low frequencies
- for tracking and rejection
2. Low loop gain at middle frequencies
where:
- for stability
R = reference input
3. Low loop gain at high frequencies
D = disturbance
- known/ unknown - for noise immunity
- random/ deterministic
N = sensor or measurement noise System Response
Output
𝑲𝑮 𝟏 𝑲𝑮 𝒚(𝒕) = 𝒚𝒕𝒓𝒂𝒏𝒔𝒊𝒆𝒏𝒕 (𝒕) + 𝒚𝒔𝒕𝒆𝒂𝒅𝒚−𝒔𝒕𝒂𝒕𝒆 (𝒕)
𝒀= 𝑹+ 𝑫− 𝑵
𝟏 + 𝑲𝑮 𝟏 + 𝑲𝑮 𝟏 + 𝑲𝑮
Tracking Error where 𝐥𝐢𝐦 𝒚𝒕𝒓𝒂𝒏𝒔𝒊𝒆𝒏𝒕 (𝒕) = 𝟎
𝒕→𝒊𝒏𝒇
𝟏 𝟏 𝟏
𝑬= 𝑹− 𝑫− 𝑵
𝟏 + 𝑲𝑮 𝟏 + 𝑲𝑮 𝟏 + 𝑲𝑮 Standard Reference Inputs
Actuator Input
𝑲 - use to take advantage of linearity and
𝑼= (𝑹 − 𝑫 − 𝑵) superposition
𝟏 + 𝑲𝑮
- simple Laplace transforms
Some Definitions: o Step input
1. Return difference 𝑳 𝟏
𝒖(𝒕) ↔
𝑱 = 𝟏 + 𝑲𝑮 𝒔
2. Sensitivity o Ramp input
𝟏 𝟏 𝑳 𝟏
𝑺= = = 𝒕𝒖(𝒕) ↔ 𝟐
𝟏+𝑲𝑮 𝑱 𝒔
3. Complementary Sensitivity o Parabolic input
𝑲𝑮 𝒕𝟐 𝑳 𝟏
𝑻=𝟏−𝑺= 𝒕𝒖(𝒕) ↔ 𝟑
𝟏 + 𝑲𝑮 𝟐 𝒔
Substituting defined variables to system variables o Sinusoidal input
1. output 𝑳 𝒔
𝒄𝒐𝒔𝝎𝒕 ↔ 𝟐
𝒀 = 𝑺𝑫 + 𝑻(𝑹 − 𝑵) 𝒔 + 𝝎𝟐
2. error 𝑳 𝝎
𝒔𝒊𝒏𝝎𝒕 ↔ 𝟐
𝑬 = 𝑺(𝑹 − 𝑫 − 𝑵) 𝒔 + 𝝎𝟐
3. input
𝑼 = 𝑲𝑺(𝑹 − 𝑫 − 𝑵) Error Response

Control System Objectives

1. Disturbance Rejection
- reduce effect of D
𝟏
- needs small S 𝑬= 𝑹
2. Good Tracking 𝟏 + 𝑯𝑮
- small E Steady-State Error
- needs small S 𝒆𝒔𝒔 = 𝐥𝐢𝐦 𝒔 𝑬(𝒔)
𝒔→𝟎
3. Noise immunity
- reducing the effect of N System Type
- needs small T, large D - no. of poles at zero
4. Bounded actuator signals
System Order
- how many poles Relationship of Td and Tr w/ system parameter a
𝟐.𝟐 𝟎.𝟔𝟗
𝑻𝒅 = = 𝟐. 𝟐𝛕 𝑻𝒓 = = 𝟎. 𝟔𝟗𝛕
𝒂 𝒂
Feedback
- allows system to react Second Order Systems
- may worsen transient response

Error Constants ???????????

Performance Specifications
1. Time Domain
a. overshoot
b. settling time
c. rise time
d. time to peak
2. Frequency Domain
a. natural frequency
b. damped frequency where:
c. damping ratio Mp = peak overshoot
d. bandwiwdth Tp = time to peak overshoot
Ts = settling time
Poles Affect Time Behavior

DC Gain
- steady-state output of the system in
response to a step input
note: some high-order systems behave like
low-order system
Poles of the second-order system
First-Order Systems

where:

System Characteristics (study the weird slide)

Rise Time Tr
- time for the step response to go from 10%
to 90%

Delay Time Td
- time for the step response to reach 50% of Step Response
the final value
Location of poles in the s-plane

REVIEW NOTES ON ONENOTE, QUIZZES then

BOOK