Beruflich Dokumente
Kultur Dokumente
Automation Lab
Experiment 1
Linear Systems
Protocol
Matlab based Linear System Analysis
2: Hamed Sepehrnia
213691
EEIT
4: Ventakesh Uppala
213961
EEIT
Contents
1 Preparation 4
2 Simulation 8
2.1 Task 4.1 - Step Response of 1st Order Systems . . . . . . . . . . . . . . . . 8
2.2 Task 4.2 - Step Response of Integrator Elements . . . . . . . . . . . . . . . 10
2.3 Task 4.3 - Step Response of 2nd Order System without Oscillations . . . . 11
2.4 Task 4.4 - Step Response of 2nd Order System with Oscillations . . . . . . 12
2.5 Task 4.5 - Frequency Response of 1st Order Systems . . . . . . . . . . . . 14
3 References 15
4 Appendix 16
1
List of Figures
2
List of Tables
3
1 Preparation
4
1 Preparation
their linear combinations yield linear systems of various types. The remaining requested
information has been provided in the appendix.
5
1 Preparation
K
G(s) = eTd s
Ts + 1
3
T = (t63% t28% )
2
Td = t63% T
For higher order systems with an S-shaped response, an n-th order system approximation
can be made with the following representation with K denoting the system gain and the
T denoting the time constant:
K
G(s) =
(T s + 1)n
Let T90% and T10% denote the times taken by the unit step response to rise to 90% and
10% of its steady state value respectively. These values can be determined from the step
response of the system. The ratio of these values then can be used to determined an
approximation of the system order using the reference table. The system order can then
in turn, be used to calculate the system time constant using the same reference table.
The integrator time is generally defined as the time taken by the integrator to generate
a response equal to the proportional component if they were part of a dependent-gain
controller. Thus, the integral time is the inverse of the integrator gain. This value can
be identified by provided a step change to the integrator. This will cause the integrator
output to rise at a fixed rate after a transient period during which its slop changes to its
steady state slope. The integrator response can then be modeled as being defined by a
gain and a dead time. The dead time can be determined by drawing an asymptote along
the the curve once it has acquired its steady state slope. The intercept on the time axis
can then be be approximated as the dead time. The integral time is the reciprocal of the
integrator gain KI which, in turn, is the ratio of the slope of the response curve and the
amplitude of the input step change.
6
1 Preparation
The systems model and parameters are specified as follows with PV and MV denoting
process variable and manipulated variable respectively:
1 Td s
G(s) = e
TI s
P V
KI =
t M V
1
TI =
KI
Question 9 - Estimating Damping and Time Constant from Overshoot and Peak
Time
The relevant parameters can be calculated using the following relations:
1
D=q
1 + ( ln (%OS/100)
)2
wn =
Tp 1 D2
1
Tn =
wn
7
2 Simulation
In this section, the lab tasks have been simulated using Matlab.
Figure 2.1: Comparison of step responses of the specified 1st order systems.
Table 2.1: Step response parameter estimates for 1st order systems.
System K T Kest Test
a 1 0.1 1 0.1012
b 1 0.5 1 0.51
c 0.5 0.1 0.5 0.1012
The time constant T determines the response time of the system as it is the time taken
by the system to attain 63% of its steady state value. A system settles to 99% of its
steady state value after a period of approximately 5 time constants. Thus, as the time
constant increases, the system will take a longer time to attain steady state. For the unit
step responses, the time constants affect the rise time and the settling time of the system
without influencing the steady state values.
8
2 Simulation
The gain K is the ratio of change in output to the change in input of a system. A large
value of K implies that a small change in input leads to a large change in output whereas
a small value of K implies that large changes in input result in small changes in output.
For unit step responses, variation in K translate directly into the steady state values of
the output responses. It does not affect the settling time or the rise time.
9
2 Simulation
The integral time TI determines the time required for the integrator element output to
generate an output that would be equivalent to that of a proportional element if the two
were in a dependent gain configuration in a controller. Thus, large values of TI imply a
slow integrator response and small values of TI translate into a fast integrator response as
evident by the slopes of the response curves of the two systems.
10
2 Simulation
The relevant values for system order and paramater estimation were calculated using a
function developed in Matlab. The relevant calculations are shown as follows:
T10% = 0.0553
T90% = 0.4006
T90%
= 7.25
T10%
T90%
= 3.89
T
T90%
T = = 0.103
3.89
Thus, the estimated time constant of the system of 0.103 is very close to the actual one
(T=0.1).
11
2 Simulation
2.4 Task 4.4 - Step Response of 2nd Order System with Oscillations
The step responses of the various system are compared in Fig. 2.4.
K D Tn
a) 1.0044 0.1033 0.4973
b) 1.0043 0.3050 0.4866
c) 1.0043 0.1033 0.9947
For the step response, the damping ratio D of the system affects the percentage overshoot
and the settling time of the system without significantly altering the rise time of the
system. Thus as the damping factor is increased, the overshoot is reduced, the system
undergoes less oscillation and settles to its steady state value more quickly.
The natural frequency influences of the system defines the frequency at which the
system would have oscillated in the undamped case and, in the damped case, affects the
damped frequency of oscillation in conjunction with the damping ratio. Thus this affects
the oscillation frequency of the response. Increasing the natural period Tn , which is the
12
2 Simulation
Case D < 0: The system will be unstable. The oscillations of the system will increase
to the bounds of the system. The poles of the system are in right hand side.
Case D = 0: The system is marginally stable. This is the undamped case. The
system output will be that of sustained oscillations with an amplitude dependent
upon the initial conditions. The poles of the system are on the imaginary axes.
Case 0 < D < 1: The system is stable. This is underdamped case. The system
output will undergo damped oscillations and the system output will eventually reach
a steady state output value. The system has complex conjugate poles.
Case D = 1: The system is stable. This is the critically damped case. The system
will reach the steady state output value with no oscillations. This is the fastest
response that system can have without oscillations. The poles are two real repeated
poles in the left hand side on the real axis.
Case D > 1: The system is stable. This is the over-damped case. The system will
reach the steady state output value relatively slowly (will consume more time than
the critically damped case). The system has two real poles at different locations in
the left hand side on the real axis.
The natural frequency wn is the frequency of oscillation of the system without damping
(D = 0). The reciprocal of the natural frequency wn is the factor Tn . The damped
frequency of oscillation is given by wd = Wn D2 1. The natural frequency wn of a
system affects the position of the poles as follows:
Case D < 0: The system will be unstable. We are not interested in the transient
response.
Case D = 0: This is the undamped case. The poles of the system are s1,2 = jwn .
Case 0 < D < 1: This is the underdamped case. The poles of the system are
s1,2 = wn D jwd .
Case D = 1: This is the critically damped case. The poles of the system are
s1,2 = wn .
Case D > 1: it is over damped case. The poles of the system are s1,2 = -wn D wd .
13
2 Simulation
The estimated values of gain and phase from the plot are as follows:
360
= t = 45.2
T
P Vpeak
KdB = 20 log = 3.0103dB
M Vpeak
14
3 References
15
4 Appendix
16
Lab1Chart
05December2016 16:08