Share Email Print
cover

Proceedings Paper

On-line milk spectrometry: analysis of bovine milk composition
Author(s): Kyle Spitzer; Rainer Kuennemeyer; Murray Woolford; Rod Claycomb
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

We present partial least squares (PLS) regressions to predict the composition of raw, unhomogenised milk using visible to near infrared spectroscopy. A total of 370 milk samples from individual quarters were collected and analysed on-line by two low cost spectrometers in the wavelength ranges 380-1100 nm and 900-1700 nm. Samples were collected from 22 Friesian, 17 Jersey, 2 Ayrshire and 3 Friesian-Jersey crossbred cows over a period of 7 consecutive days. Transmission spectra were recorded in an inline flowcell through a 0.5 mm thick milk sample. PLS models, where wavelength selection was performed using iterative PLS, were developed for fat, protein, lactose, and somatic cell content. The root mean square error of prediction (and correlation coefficient) for the nir and visible spectrometers respectively were 0.70%(0.93) and 0.91%(0.91) for fat, 0.65%(0.5) and 0.47%(0.79) for protein, 0.36%(0.49) and 0.45%(0.43) for lactose, and 0.50(0.54) and 0.48(0.51) for log10 somatic cells.

Paper Details

Date Published: 12 April 2005
PDF: 10 pages
Proc. SPIE 5852, Third International Conference on Experimental Mechanics and Third Conference of the Asian Committee on Experimental Mechanics, (12 April 2005); doi: 10.1117/12.621888
Show Author Affiliations
Kyle Spitzer, Univ. of Waikato (New Zealand)
Rainer Kuennemeyer, Univ. of Waikato (New Zealand)
Murray Woolford, Dexcel Ltd. (New Zealand)
Rod Claycomb, Sensortec Ltd. (New Zealand)


Published in SPIE Proceedings Vol. 5852:
Third International Conference on Experimental Mechanics and Third Conference of the Asian Committee on Experimental Mechanics
Chenggen Quan; Fook Siong Chau; Anand Asundi; Brian Stephen Wong; Chwee Teck Lim, Editor(s)

© SPIE. Terms of Use
Back to Top