Algorithmic support for solution of classification problems and image recognition of the studied objects by the intelligent information and measuring system

Abstract

At the manufacturing plants producing thermal insulation materials that are used in a climate of the Arctic, non-destructive testing of thermo-physical properties to ensure the quality of the products is necessary. Therefore, the use of intelligent information and measuring system with appropriate algorithmic software to control the thermo-physical properties of the objects under study is important and relevant. The aim of research is to improve the accuracy and efficiency of determining thermo-physical properties as a result of solving the classification and image recognition problems of the studied objects in production and operation to prevent deterioration of thermal insulation properties under the effect of external influential factors. Image recognition algorithms are proposed for cases when the processed information is reliable or falls into the category of fuzzy. The algorithm of decision-making in the intelligent measuring system for the choice of the non-destructive testing method of thermo-physical properties in accordance with the class of materials under study is presented. The results of experimental studies of the intelligent information and measuring system implementing the developed algorithms for image classification and recognition of the studied objects are represented, these results confirming the increase in the accuracy of non-destructive testing in thermo-physical properties of materials.