Professional Documents
Culture Documents
MODULE 7:
Analysis of Discrete (Sampled-Data) Systems
N.B.:
▪ A control system is as good as its measuring
instruments,
▪ The controller is the heart of any control system
▪ The final control elements are usually control valve,
pump, etc.
This is the device that carries out the conversion of a continuous signal to
digital form.
where:
∆𝑡 =Sampling period refers to the time interval
between successive samples.
1
𝑓𝑠 = = sampling or scan rate (cycles/time)
∆𝑡
2𝜋
𝜔𝑠 = = sampling frequency (radians/time)
∆𝑡
𝑧 = 𝑒 Δ𝑡𝑠 (7.7)
z-transform argument
z=the complex z-domain argument which is equivalent to s
z-1=reciprocal of z
z-transform operator
ℤ{. } =z-transform
ℤ−1 {. } =Inverse z-transform
Definition of inverse z-transform
−1 {
1 𝑘
𝑑𝑧
ℤ 𝑓(𝑧)} = 𝑓(𝑘) = ∫ 𝑓(𝑧)𝑧
2𝜋𝑗 𝑧
𝑐
Inverse Laplace
Transform Discretize z-Transform
f(s) f(t) f(k) f(z)
Laplace Inverse
Transform z-Transform
(1−𝑒 −∆𝑡 )𝑧
Example 4: Given 𝑔(𝑧) = (
𝑧−1)(𝑧−𝑒 −∆𝑡 )
Obtain the discrete time transfer function, g(k) using:
(a) Method of partial fractions
(b) Method of long division
Solution
(a) Resolve the given function into partial fractions, we have
(1−𝑒 −∆𝑡 )𝑧 𝑃 𝑄 𝑧 𝑧
𝑔 (𝑧 ) = (𝑧−1)(𝑧−𝑒 −∆𝑡 )
= +( = − (7.15)
(𝑧−1) 𝑧−𝑒 −∆𝑡 ) 𝑧−1 𝑧−𝑒 −∆𝑡
𝑧 𝑧
𝑔 (𝑡 ) = ℤ −1{ − −∆𝑡 } = 1 − 𝑒 −𝑡
(7.16)
𝑧−1 𝑧−𝑒
Next, we need to express g(t) in sampled-data form. Recall,
y*(t)=𝑦(0)𝛿 (𝑡) + 𝑦(Δ𝑡)𝛿 (𝑡 − Δ𝑡) + 𝑦(2Δ𝑡)𝛿 (𝑡 − 2Δ𝑡) + ⋯ (7.3)
(𝟏−𝒆−∆𝒕 )𝒛 (𝟏−𝒆−∆𝒕 )𝒛
𝒈( 𝒛 ) = ( = (7.19a)
𝒛−𝟏)(𝒛−𝒆−∆𝒕 ) 𝒛𝟐 −(𝟏+𝒆−∆𝒕 )𝒛+𝒆−∆𝒕
(1 − 𝑒 −∆𝑡 )𝑧
Thus:
𝑔(𝑧) = (1 − 𝑒 −∆𝑡 )𝑧 −1 + (1 − 𝑒 −2∆𝑡 )𝑧 −2 + (1 − 𝑒 −3∆𝑡 )𝑧 −3 + ⋯ (7.19b)
𝑔(𝑘) = ℤ−1 {𝑔(𝑧)} = (1 − 𝑒 −∆𝑡 )𝛿𝑡 + (1 − 𝑒 −2∆𝑡 )𝛿(𝑡 − 2∆𝑡) + (1 −
𝑒 −3∆𝑡 )𝛿(𝑡 − 3∆𝑡) + ⋯ (7.19c)
∞
(7.19d)
−Δ𝑡
g(k) = ∑(1 − 𝑒 ) 𝛿 (𝑡 − 𝑘Δ𝑡)
𝒌=𝟎
𝜏
𝑑𝑦
+ 𝑦(𝑡) = 𝑘𝑢(𝑡) (i) 𝑦(𝑘) + 𝑎1 𝑦(𝑘 − 1) = 𝑏1 𝑢(𝑘 − 1) (a)
𝑑𝑡
Take z-transform, we have
Take Laplace transform, we have
𝑦(𝑧) 𝑏1 𝑧 −1
𝑔 (𝑠 ) =
𝑦(𝑠)
=
𝑘
(iii) 𝑔(𝑧) = 𝑢(𝑧) = 1+𝑎1 𝑧 −1 (c)
𝑢(𝑠) 𝜏𝑠+1
𝑘
Example 7: Given the transfer function 𝑔(𝑠) = , where 𝑘 = 2 and
𝜏𝑠+1
𝜏 = 1. The exact discretization of g(s) gives the pulse transfer function
∆𝑡 ∆𝑡
𝑏1 𝑧 −1 − −
𝑔 (𝑧 −1 )
= , where 𝑎1 = −𝑒 𝜏 and 𝑏1 = 𝑘 (1 − 𝑒 𝜏 ).
1+𝑎1 𝑧 −1
(a) Obtain the transient response g(s) to a unit step input. Sketch the
response.
(b) What is the steady-state gain of 𝑔(𝑧 −1 ) with ∆𝑡 = 1?
(c) Obtain the transient response of 𝑔(𝑧 −1 ) to a unit step input.
Solution
(a) We have already obtained this in previous lecture as:
5 t
∆𝑡
−
(b) With 𝑎1 = −𝑒 𝜏 = −𝑒 −1 = −0.368
∆𝑡
−
𝑏1 = 𝑘 (1 − 𝑒 𝜏 ) = 2(1 − 𝑒 −1 ) = 2(1 − 0.368) = 1.264
𝑏1 𝑧 −1 1.264𝑧 −1
𝑔 (𝑧 −1 )
= = (7.23)
1+𝑎1 𝑧 −1 1−0.368𝑧 −1
The steady state gain of 𝑔 𝑧 ( −1 ) is the its value at z=1
−1
1.264
𝑔 (𝑧 ) = =2
𝑧=1 1 − 0.368
1.264𝑧 −1
(c) from 𝑦(𝑧 −1 )
= 𝑢(𝑧 −1 ) (7.23)
1−0.368𝑧 −1
1.264𝑧 −1 1
= −1 . (7.23b)
1−0.368𝑧 1−𝑧 −1
1.264𝑧 −1
= (7.23c)
1−1.368𝑧 −1 +0.368𝑧 −2
Using polynomial long division
𝑦(𝑧 −1 ) = 1.264𝑧 −1 + 1.729𝑧 −2 + 1.9𝑧 −3 + (7.23d)
Taking inverse z-transform, we have
𝑦(𝑘) = ℤ−1 {1.264𝑧 −1 + 1.729𝑧 −2 + 1.9𝑧 −3 +} (7.23e)
Example 8:
(a) Obtain the pulse transfer function of first order process without a
𝑘
zero-order hold. 𝑔(𝑠) =
𝜏𝑠+1
(b) Evaluate 𝑔(𝑧 −1 ) when 𝑘 = 5, 𝜏 = 2 and ∆𝑡 = 1
Solution
Step I
𝒕
𝒌 𝒌/𝝉 −
𝒈( 𝒕 ) = 𝓛−𝟏 { } = 𝓛−𝟏 { } = (𝒌/𝝉)𝒆 𝝉 (7.24)
𝝉𝒔+𝟏 𝒔+𝟏/𝝉
Step II
Recall,
∞
(7.4)
∗
𝐟 (𝐭) = ∑ 𝐟(𝒌𝚫𝒕) 𝜹(𝒕 − 𝒌𝚫𝒕)
𝒌=𝟎
∞
𝒌𝚫𝒕 (7.26)
−
𝐠(𝐤) = ∑ (𝒌/𝝉)𝒆 𝝉 𝒛 −𝒌
𝒌=𝟎
𝚫𝒕
With 𝒓 = 𝒆 , Eq.(7.26) can be written as
𝝉
∞
𝒌𝒕
( 𝝉 )(−𝒌)
(7.27)
−𝒌
𝐠 𝐍𝐇 (𝐤) = (𝒌/𝝉) ∑ 𝒆 𝒛
𝒌=𝟎
𝒌 𝟏 𝒌 𝟏
g NH (k) = [ ∆𝒕 ]= ( ) (7.28)
𝝉 −( )
𝝉 𝑧 −1
𝝉 𝟏−𝚽𝒛−𝟏
𝟏−𝒆
∆𝒕
−( )
Where 𝚽 = 𝒆 𝝉
𝒈 (𝒔 ) 𝒌
=
𝒔 𝒔(𝝉𝒔 + 𝟏)
𝒕
𝒈(𝒔) 𝒌 −
̌ (𝒕) =
𝒈 𝓛−𝟏 { } =𝓛 −𝟏 { } = 𝒌 (𝟏 − 𝒆 ) 𝝉 (7.31)
𝒔 𝒔(𝝉𝒔+𝟏)
Recall,
∞
(7.4)
∗
𝐟 (𝐭) = ∑ 𝐟(𝒌𝚫𝒕) 𝜹(𝒕 − 𝒌𝚫𝒕)
𝒌=𝟎
Discretizing Eq.7.31 gives
∞
𝒌𝚫𝒕 (7.4)
−
̌∗
𝒈 (𝐭) = ∑ 𝑲 (𝟏 − 𝒆 𝝉 ) 𝜹(𝒕 − 𝒌𝚫𝒕)
𝒌=𝟎
Recall
∞
(7.4*)
̌∗ (𝐭)} = ∑ 𝒈(𝒌𝚫𝒕)𝒛−𝒌
̌ (𝐳) = ℤ{𝒈
𝒈
𝒌=𝟎
∞ ∞
𝒌𝚫𝒕 𝚫𝒕
( 𝝉 )(−𝒌) −𝒌
(7.32)
−
̌ (𝐳) = 𝑲 ∑ (𝟏 − 𝒆
𝒈 𝝉 ) 𝒛−𝒌 = 𝑲 ∑ [𝒛−𝒌 − 𝒆 𝒛 ]
𝒌=𝟎 𝒌=𝟎
Eq.7.32 can further be written as
∞ ∞
𝚫𝒕
( 𝝉 )(−𝒌) −𝒌
(7.33)
̌ (𝐳) = 𝑲 ∑ 𝒛−𝒌 − 𝑲 ∑ 𝒆
𝒈 𝒛
𝒌=𝟎 𝒌=𝟎
N.B: The first and second series in (7.33) have been proved to converge
to (7.34) (Check Example 1 and Eq.7.26a-b)
𝟏 𝟏 (7.34)
𝐠̌(𝐳) = 𝑲 ( −𝟏
− 𝚫𝒕 )
𝟏−𝒛 −( )
𝟏−𝒆 𝝉 𝒛−𝟏
This simplifies to
𝚫𝒕
−𝟏 −( ) −𝟏
𝑲(𝟏−𝒛 )(𝟏−𝒆 𝝉 𝒛 −𝟏+𝟏)
𝒈( 𝒛 ) = 𝚫𝒕 (7.36a)
−𝟏 −( ) −𝟏
(𝟏−𝒛 )(𝟏−𝒆 𝝉 𝒛 )
𝚫𝒕
−( )
𝑲(𝟏−𝒆 𝝉 )𝒛−𝟏
𝒈( 𝒛 ) = 𝚫𝒕 (7.36b)
−( )
(𝟏−𝒆 𝝉 𝒛−𝟏 )
𝚫𝒕
−( )
Put 𝒆 𝝉 =𝚽
Obafemi Awolowo University 45
Chemical Engineering
Worked Example 9
𝑲(𝟏−𝚽)𝒛−𝟏
𝒈(𝒛) = (7.36c)
(𝟏−𝚽𝒛−𝟏 )
(b) With 𝒌 = 𝟓, 𝝉 = 𝟐 and ∆𝒕 = 𝟏
𝚽 = 𝒆−𝟎.𝟓 = 𝟎. 𝟔𝟎𝟕 (7.37)
𝟓(𝟏−𝟎.𝟔𝟎𝟕)𝒛−𝟏
𝒈(𝒛) = (7.38)
(𝟏−𝟎.𝟔𝟎𝟕𝒛−𝟏 )
𝟏.𝟗𝟔𝟕𝒛−𝟏
𝒈(𝒛) = (7.39)
𝟏−𝟎.𝟔𝟎𝟕𝒛−𝟏
𝒌
Example 10: Convert the continuous transfer function, 𝒈(𝒔) = to
𝝉𝒔+𝟏
discrete form using MATLAB.
Solution
The MATLAB function that helps in achieving this is “c2d”.
Typing “help c2d” in the MATLAB command prompt gives the
following:
Table 7.2