Creating and sharing knowledge for telecommunications

Wine classification and turbidity measurement by clustering and regression models

Duarte, D. ; Oliveira, N. D. Oliveira ; Georgieva, P. ; Nogueira, R.N. ; Bilro, L.

Wine classification and turbidity measurement by clustering and regression models, Proc Conf. on Telecommunications - ConfTele, Aveiro, Portugal, Vol. 0, pp. 0 - 0, September, 2015.

Digital Object Identifier:

Abstract
A new calculation method to determine turbidity values of wine samples of different types or colors is presented. This method is based on data obtained by multispectral turbidity polymeric optical fiber sensor for remote sensing. Prior wine classification is done with a combination of linear regression and expectation–maximization semi-supervised Gaussian mixture algorithm. A case study for white wine and rosé wine is presented. All samples were correctly classified by wine color. An estimated error rate of 9.8% was obtained for turbidity. This method can easily by adapted for other physical or chemical parameters of wine or to other type of liquids.