碩士 / 國立臺灣海洋大學 / 食品科學系 / 93 / Abstract
The purpose of this research is to describe the probability of using near infrared spectroscopy to analyze histamine spectrogram of the unit of the fish meal. 176 samples were collected and analyzed. The value through HPLC for the analysis of benzoylated amines is 10.6-109.6 ppm. There were 125 samles under 100 ppm, 44 samples between 100-500 ppm, and 7 samples beyond 500 ppm. WinISI Ⅱwas used to analyze the correlation between histamine contents and absorption spectrums of fish meal measured by near infrared spectroscopy. The 2nd derivative treatments of absorption spectrum have more information and the absolute value of correlation increased to more than 0.5.
The Modified Partial Least Square Regression (MPLSR), Principal Component Regression (PCR) and Artificial Neural Network (ANN) were used in discussion in the whole bands of 400-1098 and 1100-2498, and the partial bands of 700-900, 1200-1300, 1600-1700, and 2000-2200 under different mathematic models (original, first derivative and second derivative) and factors to find out the best prediction model. The MPLSR shows the best result in linear regression partial bands (factor = 5, rc =0.95, SEC=22.67 ppm, rp =0.85, SEP=36.21, RPD=1.88).
In non-linear model, ANN shows the best result in both bands of study. (In the whole bands: rc =0.99, SEC=9.24 ppm, rp =0.98, SEP=12.69 ppm, RPD=5.51; in the partial band: best factor is 3 , rc=0.97, SEC=16.57 ppm, rp =0.88, SEP=33.72 ppm, RPD=3.06).
The best prediction model, ANN shows the relationship between the values of prediction and chemical analysis. There are 53 samples whose contents of histamines are under 100 ppm ( rc =0.56, SEC=20.48 ppm ). There are 16 samples in another group, whose contents of histamine are between 100 and 350 ppm. The relative coefficient between the values of prediction and chemical analysis are rc =0.95, SEC=13.87 ppm. The results show the groups with higher concentration have higher relationship.
| Identifer | oai:union.ndltd.org:TW/093NTOU5253071 |
| Date | January 2005 |
| Creators | Shu-chen Kuo, 郭淑貞 |
| Contributors | Tze-Kuei Chiou, 邱思魁 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 111 |
Page generated in 0.0083 seconds