The research of technology of apple extraction of a certain form in the conditions of the food industry with application of clusterization methods

Tsvirkun, Liudmyla and Цвіркун, Людмила Олександрівна and Цвиркун, Людмила Александровна and Цвіркун, Сергій Леонідович and Tsvirkun, Sergij and Цвиркун, Сергей Леонидович and Бондаренко, Олена Олександрівна and Бондаренко, Олена Олександрівна and Bondarenko, Olena (2020) The research of technology of apple extraction of a certain form in the conditions of the food industry with application of clusterization methods. Обладнання та технології харчових виробництв (41). pp. 97-104. ISSN 2079-4827

Preview

Text Tsvirkun_article_10_02_2021+.pdf Download (1MB) | Preview

Abstract

It is noted that at the preliminary stage of sorting there is a removal of small apples, broken, low-quality, as well as with defects; by size, variety, color apples are sorted later in production. Emphasis is placed on the fact that the affiliation of apples of a certain species can be determined by several parameters, namely size (d), weight (m), color (g), so it is advisable to use the clustering operation to perform this operation. It is proposed to use clear and fuzzy clustering methods to divide apples into certain varieties represented by several varieties. Analysis of research has shown that a clear clustering of the characteristics of apples X means the division of data into a certain number of mutually exclusive subsets with similar characteristics that are inherent in a particular variety. In this case, it is believed that the number of species is known. Using classical set theory, a clear clustering is defined as a family of subsets corresponding to the conditions: all objects are distributed in clusters, each object belongs to only one cluster none of the clusters is empty. It was found that among the methods of clear clustering to solve the problem of recognizing apples by variety, the most effective methods are K-means and K-medoid, which determine the affiliation of each set of characteristics of apples for one of the clusters. The results of fuzzy clustering of apples by algorithms (Fuzzy C-means, Gustafson-Kessel, Gus Giva) are interpreted. The results of fuzzy partitioning in the form of a three-dimensional graph are presented. It is proposed to divide apples into certain varieties represented by several varieties, the similarity of which will exclude a certain type of apples from the general flow, a method that allows the recognition of certain varieties of apples by fuzzy clustering of its characteristics using Gustafson-Kessel algorithm and subsequent approximation clustering to multiple inputs.

Actions (login required)

View Item