Artificial neural network as a basic element of the automated goniometric system

Abstract

The approach to automatic definition of components of the systematic error and the sources of its appearing in the automated goniometric system that is based on artificial neural networks are proposed in the article. In particular, the input and output vectors and the structure of artificial neural network are defined. For this propose systematic error of automated goniometric system is presented as a totality of instrumental, methodic and subjective components, and each of them has defined primary components. These components form the structure and content of the artificial neural network input vector. The structure and the content of the output vector allow to detect the causes of errors and to correct measuring result in future. Generalized methodics of the proposed “back-propagation” neural network is given. The last one will be trained by the supervised learning.