Please use this identifier to cite or link to this item:
https://ptsldigital.ukm.my/jspui/handle/123456789/457582Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | Zahira Yaakob, Prof. Ir. Dr. | - |
| dc.contributor.author | Liew Weng Hui (P52847) | - |
| dc.date.accessioned | 2023-09-12T09:11:11Z | - |
| dc.date.available | 2023-09-12T09:11:11Z | - |
| dc.date.issued | 2013-02-28 | - |
| dc.identifier.other | ukmvital:74983 | - |
| dc.identifier.uri | https://ptsldigital.ukm.my/jspui/handle/123456789/457582 | - |
| dc.description | Biodiesel emerges as a highly renewable and environmental-friendly fuel to support the world energy demand in future. In order to increase the competitiveness against fossil fuel, maximum process efficiency is important to reduce the production cost. Process modeling for understanding of process characteristics is recognized as an important step for process optimization and control. However, conventional modeling method poses shortfall since it incurs more effort and time for determining the value of model coefficients particularly when the model structure is complex. In this study, artificial neural network (ANN) as advanced modeling tool was used to model the biodiesel production through transesterification process. The ANN consists of the simplest structure with an input layer, a single hidden layer and an output layer. Amount of input neurons depends on the number of input data sets while output neurons refer to the number of predicted outputs. The training of ANN is executed by Levenberg-Marquardt (LM) algorithm for rapid error convergence. The ANN simulation program with effective training features was self-developed using Matlab® software (Version 7.04, U.S.A.). The training sequences were configured to have the features of continuous error reduction in training phase and cross-assessment in testing and validation phase. Detailed optimization of ANN was performed by manipulating the internal training parameters, which includes the number of hidden neurons, initial range of weights and biases and damping factor in LM algorithm. To illustrate the ANN optimization result and modeling capability, three case studies are presented and discussed. Those studies assess the ANN performance under different type of data (discrete and continuous) and transesterification process (catalyzed and enzymatic type). As results, ANN with low number of hidden neurons was able to generate the best result in terms of mean squared error (MSE), correlation coefficient (R) and number of epochs. For Case Study 1 on catalyzed transesterification process with continuous data, ANN with 5 hidden neurons generated the MSE of 4.4500 x 10 -6 , R of 0.9998 and 16 epochs. Case Study 2 refers to the modeling of discrete data under catalyzed transesterification and the results achieved were MSE of 3.5116 x 10 -3 , R of 1.0000 and 28 epochs. In Case Study 3, ANN was applied to enzymatic transesterification process with continuous data and the modeling was based on two different models of sunflower and soybean oil. For sunflower oil, the results were MSE of 9.4000 x 10 -7 , R of 0.9999 and 21 epochs; while for soybean oil, the results were MSE of 1.0000 x 10 -8 , R of 1.0000 and 9 epochs. ANN response is sufficiently robust in modeling the data with noise. Enlarging the initial range of weights and biases did not generate significant improvement to the performance. The proposed updating method of damping factor in LM algorithm had improved the number of epochs and the simulation duration. Under comprehensive study based on the type of data and process, the optimized ANN is potentially capable of modeling the transesterification process for biodiesel production.,Master | - |
| dc.language.iso | eng | - |
| dc.publisher | UKM, Bangi | - |
| dc.relation | Faculty of Engineering and Built Environment / Fakulti Kejuruteraan dan Alam Bina | - |
| dc.rights | UKM | - |
| dc.subject | Artificial Neural Network Modeling | - |
| dc.subject | Artificial Neural Network Modeling On Biodiesel Production | - |
| dc.subject | Neural Network Modeling On Biodiesel Production | - |
| dc.subject | Artificial Neural Network Modeling On Biodiesel Production Through Transesterification Process | - |
| dc.subject | Biodiesel fuels | - |
| dc.title | Artificial Neural Network Modeling On Biodiesel Production Through Transesterification Process | - |
| dc.type | theses | - |
| dc.format.pages | 170 | - |
| dc.identifier.callno | TP359.B46 .L835 2013 3 | - |
| dc.identifier.barcode | 000467 | - |
| Appears in Collections: | Faculty of Engineering and Built Environment / Fakulti Kejuruteraan dan Alam Bina | |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| ukmvital_74983+Source01+Source010.PDF Restricted Access | 1.74 MB | Adobe PDF |  View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.