This PhD thesis reports on the potential application of an electronic nose to analysing the quality of potable water. The enrichment of water by toxic cyanobacteria is fast becoming a severe problem in the quality of water and a common source of environmental odour pollution. Thus, of particular interest is the classification and early warning of toxic cyanobacteria in water. This research reports upon the first attempt to identify electronically cyanobacteria in water. The measurement system comprises a Cellfacts instrument and a Warwick e-nose specially constructed for the testing of the cyanobacteria in water. The Warwick e- nose employed an array of six commercial odour sensors and was set-up to monitor not only the different strains, but also the growth phases, of cyanobacteria. A series of experiments was carried out to analyse the nature of two closely related strains of cyanobacteria, Microcystis aeruginosa PCC 7806 which produces a toxin and PCC 7941 that does not. Several pre-processing techniques were explored in order to remove the noise factor associated with running the electronic nose in ambient air, and the normalised fractional difference method was found to give the best PCA plot. Three supervised neural networks, MLP, LVQ and Fuzzy ARTMAP, were used and compared for the classification of both two strains and four different growth phases of cyanobacteria (lag, growth, stationary and late stationary). The optimal MLP network was found to classify correctly 97.1 % of unknown non-toxic and 100 % of unknown toxic cyanobacteria. The optimal LVQ and Fuzzy ARTMAP algorithms were able to classify 100% of both strains of cyanobacteria. The accuracy of MLP, LVQ and Fuzzy ARTMAP algorithms with 4 different growth phases of toxic cyanobacteria was 92.3 %, 95.1 % and 92.3 %, respectively. A hybrid e-nose system based on 6 MOS, 6 CP, 2 temperature sensors, 1 humidity sensor and 2 flow sensors was finally developed. Using the hybrid system, data were gathered on six different cyanobacteria cultures for the classification of growth phase. The hybrid resistive nose showed high resolving power to discriminate six growth stages as well as three growth phases. Even though time did not permit many series of the continuous monitoring, because of the relatively long life span (30-40 days) of cyanobacteria, improved results indicate the use of a hybrid nose. The HP 4440 chemical sensor was also used for the discrimination of six different cyanobacteria samples and the comparison with the electronic nose. The hybrid resistive nose based on 6 MOS and 6 CP showed a better resolving power to discriminate six growth stages as well as three growth phases than the HP 4440 chemical sensor. Although the mass analyser detects individual volatile chemicals accurately, it proves no indication of whether the volatile is an odour. The results demonstrate that it is possible to apply the e-nose system for monitoring the quality of potable water. It would be expected that the hybrid e-nose could be applicable to a large number of applications in health and safety with a greater flexibility.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:343213 |
| Date | January 1999 |
| Creators | Shin, Hyun Woo |
| Publisher | University of Warwick |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://wrap.warwick.ac.uk/111031/ |
Page generated in 0.0173 seconds