Application of multifractal wavelet analysis to spontaneous fermentation processes
Rosu Barbus, Haret-Codratian
An algorithm is presented here to get more detailed information, of mixed cul-ture type, based exclusively on the biomass concentration data for fermentation processes. The analysis is performed with only the on-line measurements of the re-dox potential being available. It is a two-step procedure which includes an Artiﬁcial Neural Network (ANN) that relates the redox potential to the biomass concen-trations in the ﬁrst step. Next, a multifractal wavelet analysis is performed using the biomass estimates of the process. In this context, our results show that the redox potential is a valuable indicator of microorganism metabolic activity during the spontaneous fermentation. In this paper, the detailed design of the multifractal wavelet analysis is presented, as well as its direct experimental application at the laboratory level.