Professional Documents
Culture Documents
MPC Fazal+Hamza
MPC Fazal+Hamza
M o d e l P r e d i c ti v e C o n t r o l
( C o u p l e d Ta n k S y s t e m )
Co u r s e : Pro ce s s Mo d e l i n g a n d Co nt ro l .
Presented by;
Fazal-e-Rabbi (MS-Process Engineering)
H a m z a I m ti a z B u tt ( M S - P r o c e s s E n g i n e e r i n g )
2
Coupled Tank System (CTS)
Non-Linear
Linearization Discretizatio MPC Design
Dynamic
using Taylor n and State and Cost
Model of
Series. Space Model. Function.
System.
Non-Linear Dynamic Model of CTS 5
Tank 1: Tank 2:
_________ (1)
____ (2)
= Fluid Velocity at Pump outlet (Manipulated Variable), = Pump outlet cross-sectional area.
= Level in tank 1, = Orifice area below tank 1, = Area of tank 1.
= Level in tank 2 (Control Variable), = Orifice area below tank 2, = Area of tank 2.
6
Model Simplification
Renaming the coefficients for simplicity:
• Declaring:
,
,
Where and are steady state operating values for tank 1 and 2.
8
Model Linearization (Cont.)
Now eq (3) and (4) become:
(5)
(6)
(7)
(8)
9
Model Linearization (Cont.)
Subtracting (7) from (5) and (8) from (6). Also writing the equation in
deviation variable form:
’=- ,’=- ,’=-
(9)
(10)
10
State Space Representation (MIMO)
General form:
𝑑𝑥
= 𝐴( 𝑥 + 𝐴
−1
𝐵
𝑑𝑡
[ 𝑙1
𝑙2 ]
= [ 1
0
0
1 ][ 𝑙1
𝑙2 ]
11
Continuous to Discrete Time State Space Model
Using the strategy from Cameron (Lemma 11.1.1):
k= Previous time step.
k+1= Future time step.
h= Sampling time.
𝑙𝑛
| 𝑥 ( 𝑘 +1 )
𝑥 (𝑘)+ 𝐴
−1
𝐵𝑢 ( 𝑘 ) |
=𝐴h
𝑥 ( 𝑘+ 1 ) = [ 𝑥 ( 𝑘 ) + 𝐴 −1
𝐵𝑢 ( 𝑘 ) ] 𝑒 𝐴h
12
Discrete-time State Space Model
Resolving the exponential by Taylor Series:
2
𝐴h ( h𝐴 )
𝑒 = 𝐼 + h𝐴 + +…
2!
𝑥 ( 𝑘+ 1 )= [ 𝑥 ( 𝑘 ) + 𝐴 −1
𝐵𝑢 ( 𝑘 ) ] ( 𝐼 +h𝐴 )
13
State Space Representation (MIMO)
General form:
(State Vector)
(Measurement Vector)
(Input Vector)
14
State Space Representation (Cont.)
Coefficient Matrices:
15
MPC Design Strategy
(A)
17
Connecting Output to State Variable Change
Defining new state variable vector and output at different time steps:
𝑥 ( 𝑘 )=
[ ∆ 𝑥𝑚 ( 𝑘 )
𝑦 (𝑘) ]
(B)
18
Augmented Model
Combining equations (A) and (B) in the new state variable :
Here:
19
Upcoming Presentations
MPC
Control
Parameter
s
Cost
Function
MATLAB /
SIMULINK
20
o u !
n k Y s ?
T h a t i on
u e s
n y Q
A