Tetyana Baydyk, Ernst Kussul, Gengis Kanhg Toledo-Ramirez
Object recognition, computer vision, LIRA neural classifier, PCNC, solar concentrators, flat mirrors
The aim of this article is to compare two neural classifiers applied to micro component recognition. These micro components can be used for the automation of production and assembly processes of solar concentrators. The image databases have been developed and include mixed and heaped-up micro work pieces. Two neural classifiers, the limited receptive area classifier (LIRA) and the permutation coding neural classifier (PCNC), were tested using these databases. Both methods have demonstrated good results. The LIRA neural classifier achieves a recognition rate of 90%, while the PCNC achieves 91%. Parameters of reliability, adjustability and other characteristics of the classifiers are compared. In the future, these algorithms can be used for the intelligent automation of renewable energy technology production. For example, we intend to use the algorithms for the assembly task of solar concentrators developed in our laboratory.
Cite this paper
Tetyana Baydyk, Ernst Kussul, Gengis Kanhg Toledo-Ramirez. (2017) Investigation of Neural Classifiers for Recognition of Micro Components of Solar Concentrators. International Journal of Mechanical Engineering, 2, 19-27