An artificial neural network approach to prediction of sugar beet yield and quality in Serbia

Abstract

Decision-making processes in agriculture often require reliable crop response models. Prediction of crop yield mainly strategic plants such as, wheat, corn and rice has since long been an interesting research area to agro meteorologists, as it is important in national and international economic programming. The main purpose of such studies is to estimate the crop production before harvesting, using meteorological data. Recently, the application of artificial neural networks has developed into a powerful tool that can compute most complicated equations and numerical analyses to the best approximation. The goal of this study was to apply the artificial neural networks to predict yield as well as the quality of sugar beet grown in Serbia. According to the available data and information such as field-specific rainfall data and the weather variables, for different areas in Vojvodina, the main agricultural region of the country, models were developed using historical yield data and quality parameters at multiple locations throughout Vojvodina. Adjusting artificial neural networks parameters such as learning rate and number of hidden nodes affected the accuracy of sugar beet yield predictions. Results suggested that artificial neural networks can be used for prediction of sugar beet yield in this region. Neural networks were less accurate when predicting sugar beet quality parameters at the province level, while at local levels predictions were more accurate. © 2010 by Nova Science Publishers, Inc. All rights reserved.