Analysis and Modelling of the Relationship between Meteorological Course and Mass of Soil Wastes in the Sugar-Beet Industry

Abstract

this work illustrates the results obtained with the application of the artificial neural network (ANN) models for the prediction of soil tare levels of sugar-beet in the course of the harvesting season on the basis of the meteorological patterns (here mainly described by the rainfall). As known, the soil tare affects the costs and the impact on the environment of the sugar production process. Numerous factors are related with the soil tare and their reciprocal relationships are complex. Among the available ANN architectures, the Elman-Jordan and General Regression Neural Network were applied to the 1997 data, collected in the central part of Italy. The results show that the values predicted by the two models were reasonably similar to the real values. However, in the cases where the variability of the soil tare were very high, the chosen models show a lower accuracy.