SYSTEMATIC DESIGN OF SIMPLY STRUCTURED COMPENSATOR

PID controllers have been the most common type of compensators in the past and will
continue in the near future. Effort has been taken by researchers to develop simple
and systematic way to design these controllers.
This project aims to develop the algorithm for the tuning method that based on
Nyquist Stability Criterion and at later stage build a Neural Network Model to predict
the tuning parameters for the PID controller.
Recently, it has been revealed the possibility that the fine-tuning method based on
Nyquist Stability Criterion has the potential to output a satisfactory closed-loop
performance for a given continuous, linear system model. In addition, design of
compensator based on Nyquist Stability Criterion is simpler and more robust [3]. On
the other hand, Neural Network has become tremendously popular in the control
application due to its ability in adaptive learning and approximating function. By
implementing Nyquist Stability Criterion's tuning algorithm with Neural Network,
this will definitely enhance the process of tuning the PID controller.
In this project, the algorithm of the tuning method based on Nyquist Stability
Criterion is developed first. Then, data are generated to be used in training the neural
network. The NN is used to predict the tuning parameters of PID controller for
second order system. The whole project is implemented in MATLAB program.
The result obtained in this project has actually showed that the tuning rule based on
Nyquist Stability Criterion can fine-tune the PID controller and eventually improve
the closed-loop system. Meanwhile the NN model built has been able to predict the
parameters for PID controllerat the accuracyof 5%MSE.
In the final part of this report, conclusions are drawn and some future work for future
development is proposed.