NEURO-FUZZY LOGIC CONTROLLER FOR HEAT EXCHANGER TEMPERATURE CONTROL

A commonly used controller in the process industries is the Proportional-Integral-
Derivative (PID) controller due to its features; cheap and easy to configured. Another
type of controller that is now being developed is neuro-ruzzy logic controller that
functions like a human brain which consists of interconnected processing elements
called nodes or neurons that work together to produce an output function. The feature
that it has that the PID controller does not have is the ability to be retrained to deal with
various conditions in the process industries. A previous final year project on neuroruzzy
logic controller yielded an unsatisfactory result as it could only perform for a
single set point change [1]. The objective of this project is to improve the neuro-ruzzy
logic controller so that it can control a process for a wider range of set point values.
Data achieved for this project are through plant experiments using SIM 305 Pilot Plant:
Plant 6 for the purpose of process modeling and computer simulation. Computer
simulation is used to design the PID controller and the neuro-fuzzy logic controller.
These two types of controllers are then compared and analyzed based on their
performances. Controlled Variable (CV) Overshoot, Manipulated Variable (MV)
Overshootand DecayRatio are the benchmarks used to compare and evaluate these two
controllers. Based on the simulation results, the two controllers are in par since the
benchmark values for the two controllers are nearly the same. However, it can be
concluded at this stage that the PID controller is better than the neuro-fuzzy logic
controller due to the smallest value of overshoot compared to the neuro-fuzzy logic
controller.