Sie sind auf Seite 1von 9

Part 1: Evaluating different controllers

In this part of our experiment we examine performance of different forms of the PID
controller. The plant consists of simple mass-spring-damper system.

Task 5.3
Initially we check the performance of the P-controller with Kp=1.

Figure
5.1:
Step
response
of the P-

controller(Kp=1)
We see that the measured speed never reaches setpoint. That is because with only P-
controller we have always steady-state error.

Task 5.4
From Root Locus we found that Kp=2.58 is stability limit.

Figure
5.2:
Step
response
of the P-
controller(Kp=2.58)
We get oscillations around setpoint with constant amplitude. So it means that we are
exact at the stability limit.

Task 5.5
In this task we increase proportional gain step by step and study the behavior of the
system.

Figure
5.2:
Step
response
of the P-

controller(Kp=0.5)

Figure
5.3:
Step
response
of the P-
controller(Kp=1.5)

Figure
5.4:
Step
response
of the P-

controller(Kp=2)
With increasing Kp we get less steady-state error. It is not possible bring system to the
steady-state with P-only controller. However if we make gain too high it become
unstable. From previous task we found that Kp=2.58 is the stability limit.

Task 5.6
We have already found value of gain for the stability limit in task 5.4. In stability limit
periodic oscillations with constant amplitude around setpoint will appear.

Figure
5.5:
Response in the stability limit

Task 5.7
In this task we add integral part and study PI-controller with parameters: Kp=1, TI=10.

Figure
5. 6:
Step
response
of the
PI-

Controller
Now we see that it is able to get steady-state value because of integral part. Also we
see around 80% overshoot.

Task 5.8
We change integral time of PI-controller and will see how it affects to the response.

Figure
5.7:
Step
response
of the
PI-
controller(Kp=1, TI=15)

Figure
5.8:
Step
response
of the
PI-

controller(Kp=1, TI=30)

Figure
5. 9:
Step
response
of the
PI-

controller(Kp=1, TI=70)
Figure
5.10:
Step
response
of the
PI-

controller(Kp=1, TI=200)
By increasing integral coefficient we get smaller overshoot, but control loop gets
slower. With sufficient enough integral part we get no more overshoot. From our plots
we can predict this value for TI is between 70 and 200. Also we have more vibrations
with increasing integral part.

Task 5.9
In this task we keep constant TI=90 and vary value of proportional gain.

Figure
5.11:
Step
response
of the
PI-

controller(Kp=0.01, TI=90)
Figure
5.12:
Step
response
of the
PI-

controller(Kp=0.1, TI=90)

Figure
5.13:
Step

response of the PI-controller(Kp=1, TI=90)


We can see that we very small Kp system does not reach its steady state in given time.
By increasing gain we get faster response, but after some value of gain response will
overshoot.
Task 5.10
Now we can state which values of controller parameters are best suits to our
requirements. The chosen coefficients: Kp=0.17, TI=90.

Figure
5.14:
Step
response
of the
PI-

controller
We see that chosen parameters brings satisfactory outcome. We have almost no
overshoot and vibrations. Also rise time is also satisfactory.

Task 5.11
In this task we study the performance of PD-controller. The parameters of the
controller: Kp=1, TD=0.1.

Figure
5.15:
Step
response
of the
PD-
controller
The system is not able to reach steady state, because we have no integral part. The rise
time is less because of presence differential part, but it creates output vibrations
during rise time. Also initial big change of setpoint brings big jump of the output.

Task 5.12
We see that without integral part PD-controller can not eliminate stead-state error.
Also we will from suffer big output jumps during big change in setpoint. For this
reason it is convenient to keep small value of differential gain if needed. We can
conclude that response to the step change is not acceptable with PD-controller.

Task 5.13
Basing in our experience from previous tasks now we chose appropriate values for
PID-controller: Kp=0.1, TI=75 and TD=5.

Figure
5. 16:
Step
response
of the
PID

controller
The response has no overshoot and has sufficient rise time. Also we don’t see any big
vibrations and rapid jumps. We can say that PID controller with chosen parameters
has good performance.

Das könnte Ihnen auch gefallen