Study on Neural Network Predictive Controller: Impact ofNetwork's Architecture and Plant-Model Mismatch

This final report explains about the extended research done on basic concept of the
selected topic, which is Study on Neural Network Predictive Controller: Impact of
Network's Architecture and Plant-model mismatch. The objective of the project is to
study the effect ofdifferent network's architecture by manipulating the transfer fiinction,
number ofneurons, weight and biases on NN-based Predictive Controller's performance.
And to study the impact of parameters used in CSTR on the performance of the based
model in Plant-Model Mismatch. In literature review section, NN-based Predictive
Controller will be further discussed. The successful outcomes of this project can be
applied in the industries in orderto help reducing the uncertainty or inaccuracy in process
control.