Identification and modelling of two phase dc-dc boost converter based on autoregressive moving average with exogenous, output-error and transfer function model structures
This research presents the identification and modelling of a two-phase DC-DC boost converter based on the autoregressive moving average with exogenous (ARMAX), output-error (OE) and transfer function (TF) model structures for low-voltage applications. The goals that led to this study were to r...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/8280/1/24p%20MOHD%20AZLEE%20NOOR%20AMRAN.pdf http://eprints.uthm.edu.my/8280/2/MOHD%20AZLEE%20NOOR%20AMRAN%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/8280/3/MOHD%20AZLEE%20NOOR%20AMRAN%20WATERMARK.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research presents the identification and modelling of a two-phase DC-DC boost
converter based on the autoregressive moving average with exogenous (ARMAX),
output-error (OE) and transfer function (TF) model structures for low-voltage
applications. The goals that led to this study were to reduce the time taken to design
the controller and analyse the output of constant Kp and Ki generated from the auto
tuning method. A two-phase boost converter employs as 180-degree phase shift from
each phase to drive the power switch. This research focused more on the system
identification approach to generate mathematical models from the open-loop
response. The generated models were from the TF, ARMAX and OE model
structures. The mathematical models were generated from the pulse-width
modulation (PWM) input and voltage output of the two-phase boost converter itself
in the time domain data. After the best model order was found to replace the two�phase boost converter with a mathematical model, the controller design took place.
Some closed-loop blocks were designed for the mathematical models in
MATLAB/Simulink software, which were also used to perform the auto-tuning of
the proportional-integral (PI) controller. However, tuning methods such as the
Ziegler-Nichols and the Cohen-Coon methods are more time-consuming. After the
best values for constants Kp and Ki were determined, the values were used in the real
hardware to analyse the output responses. The findings showed that Kp and Ki from
the TF model showed 19% overshoot compared with those of the ARMAX and OE
models, which were 25.36% and 24.6%, respectively. All of the output responses
from the different Kp and Ki values resulted in less than 5% ripple voltage. It can be
concluded that the best model from the system identification approach was the TF
system model, since it had the lowest overshoot and the lowest percentage of output
voltage ripple |
|---|