The identification of relationships between selected honey parameters by extracting the contribution of independent variables in a neural network model

Abstract

Honey is a product having many valuable properties for human health. The honey quality control regarding its floral or geographical origin, as well as a potential adulteration, should be carried out at any stage of processing and marketing. At present, some new methods of honey quality assessment based on its electrical parameters measurement are developed. The development of accurate methods, which could be useful in practice, requires the knowledge of the relationships between honey chemical and electrical parameters. In particular, the level of the influence of chemical parameters which are modified during honey quality deterioration on electrical features is of the great importance. When the complex impedance is used as electrical parameter, the influence of frequency on relationships investigated is essential. In this work, the influence of glucose content/fructose content ratio, water activity, HMF content, pH, diastase as well as the temperature on honey impedance measured in the frequency of 260 Hz, 11 kHz and 160 kHz has been investigated. Three different methods of the determination of independent variables contribution in a neural network model have been employed. The highest influence on the real part of impedance measured for each frequency has been calculated for HMF content and pH. Similar results have been obtained in the case of the imaginary part of impedance measured for the frequency of 260 Hz. For higher frequencies, the increase in the influence of water activity, diastase and glucose content/fructose content ratio on the imaginary part of impedance has been observed.