Development of a predictive tissue discolouration model based on electronic potato impacts

Farmers, processors and manufacturers of potato-handling equipment have a strong need for fast support systems to analyse and to evaluate the bruising risk while handling potatoes. The abundance of influencing factors, their interactions and the complex relationships between mechanical impacts and the resulting tissue discolouration, however, hamper an easy solution. The main objectives of this work were the introduction and construction of a frame of reference for the PTR 200 electronic potato and the development of a predictive tissue discolouration model taking into account both impact energy level and a reduced number of potato bruise susceptibility parameters. The attention was mainly focussed on dry matter content and tuber temperature as influencing factors because of their fast determination possibilities, which is essential for an online evaluation and support system. In the 2000, 2001 and 2002 harvesting seasons, field experiments were carried out with seven potato varieties on harvesters and storage lines using the PTR 200 electronic potato. For each harvester or storing line the individual energy levels of the impacts were measured and the values were used to compute a general energy level, set as a sensor index Is. Produce samples were taken at several points in the processing chain and analysed for tissue discolouration and dry matter content. Both characterisations resulted in average values for energy level, discolouration and dry matter content. Tuber temperature, cultivar differences and-in some cases-potassium (K) content of the tubers were taken into account. Multiple linear regressions were performed and analysed. The regression models showed moderate coefficients of determination. On the basis of the statistical models, discolouration prediction intervals were computed for new measurements with the PTR 200. This allowed the prediction of tissue discolouration in potato by means of PTR 200 measurement results. Due to the limited number of bruise susceptibility factors in the models, the seasonal and biological variability and the moderate coefficients of determination, prediction intervals are rather wide which limits the predictive power of the models. (C) 2004 Silsoe Research Institute. All rights reserved Published by Elsevier Ltd.