Modeling Of Biopolymerization Process Using First Principle Model And Bootstrap Re-Sampling Neural Network

Simple item page
dc.contributor.authorMat Noor, Rabiatul 'Adawiah
dc.date.accessioned2018-08-30T02:49:57Z
dc.date.available2018-08-30T02:49:57Z
dc.date.issued2011
dc.description.abstractThe emergence of the environmental issues such as green house and global warming have triggered scientists and researchers to create new materials that can cope both environmental and humanity needs i.e. biopolymer. Biopolymer quality assesses by its molecular weight. Apparently, there is no online measurement for molecular weight measurement. Therefore, in this study models are developed to mimic the real process using a reliable modeling tool such as neural networks. First principle model also become one of the most applied methods to model a process other than using a modeling tool such as neural network. Therefore in this research, neural network and first principle model have been chosen to model the biopolymerization process and followed by the comparison between the aforementioned models. In order to develop the models, experimental work is conducted to obtain data for first principle model and neural network model. The data for the neural network model are the molecular weight of the biopolymer whereas the data for the first principle model are the kinetics data. Both models delivered predicted molecular weight of biopolymer using different approaches i.e. fundamental and empirical model. Both models delivered convincing results in terms of molecular weight number as well as molecular weight trend. Based on the results from both models, neural network model gives closest prediction with sum squared error (SSE) is 0.9996 and correlation coefficient, R value is 0.9999. First principle model results on the hand assessed based on Analysis of Variance (ANOVA). In this work, ANOVA takes the significant level, α as 0.01. Hence, the results that give the value that is less that 0.01, is showing that the compared data are significantly different from each other which is undesirable as the actual and predicted data should be projected similar trend and number.en_US
dc.identifier.urihttp://hdl.handle.net/123456789/6480
dc.language.isoenen_US
dc.publisherUniversiti Sains Malaysiaen_US
dc.subjectModeling of biopolymerization process using first principle modelen_US
dc.subjectand bootstrap re-sampling neural networken_US
dc.titleModeling Of Biopolymerization Process Using First Principle Model And Bootstrap Re-Sampling Neural Networken_US
dc.typeThesisen_US
Files
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Download
Collections
Pusat Pengajian Kejuruteraan Kimia - Tesis