Offie, Ikechukwu Chukwudi (2023) Development of Artificial Intelligence Systems for Anaerobic Digestion Operations. Doctoral thesis, University of West London.
Preview |
PDF
Offie I - PhD Thesis Final April 24.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (3MB) | Preview |
Abstract
This study explores two novel approaches for improving the performance of a micro anaerobic digestion system in generating maximum biogas. The micro anaerobic digestion system was a wet system situated in Camley-Central London. It operated continuously for 310 days under mes�ophilic conditions. The novel approaches include a new artificial intelligence-based model frame�work and an ensemble-based model framework. Both frameworks were developed using historic
data obtained from the micro- anaerobic digestion system. The historic data include feed, cater�ing, oats, liner, water, and biogas. The new artificial intelligence-based model framework entails developing a Recurrent Neural Network model for predicting biogas generated from the micro�anaerobic digestion system. The ensemble-based model framework entails combining different weak learning data mining models to improve the prediction accuracy of biogas generated. These weak learning data mining models include Support Vector Machines, K-Nearest Neighbour, De�cision Tree, Gaussian Process Regression, Discriminant Analysis and Naïve Bayes. Both models were optimised after being trained to predict biogas using shuffled frog leaping algorithm to obtain the maximum biogas volume. The results showed great potential for the developed new artificial intelligence-based model in improving the performance of the micro anaerobic system in yielding optimal biogas by 43%. The results also showed that the average biogas produced could increase from 3.26 to 4.34 m3/day. The developed ensemble model demonstrated 91% biogas prediction accuracy from the micro- anaerobic digestion system. The results of the weekly operation pattern led to 78% increase in biogas generation during the testing period. It also contributed to a 71% reduction in total required feeding days and 30% reduction in required pre-feeding days. The novel approaches demonstrated promising potentials in improving the performance of the micro�anaerobic digestion system to obtain maximum biogas with minimum energy and low operational costs making it a more viable option for managing organic wastes.
Item Type: | Thesis (Doctoral) |
---|---|
Identifier: | 10.36828/xvqy2136 |
Subjects: | Computing > Intelligent systems |
Depositing User: | Marc Forster |
Date Deposited: | 04 Jul 2024 08:30 |
Last Modified: | 04 Jul 2024 08:30 |
URI: | https://repository.uwl.ac.uk/id/eprint/12136 |
Downloads
Downloads per month over past year
Actions (login required)
View Item |