New Possibilities in Using of Neural Networks Library for Material Defect Detection Diagnosis

Abstract

Use of neural network as a tool for a data selection or data mining is well known and discussed many times. Use of neural networks and new software libraries for diagnostics is however not widely used. Our described project deal with a phenomenon of new possibilities in using neural networks implemented in software library for material defect diagnostics which can save a huge amount of money in several special cases. This paper describe also a design and realization of a test application developed in object orientated programming language C#. Developed application classifies data obtained from industrial processes of metallurgical plant. Data used in this project was received during measurement of material diagnostics and its structural defects. Core of developed application is based on neural networks design, which is capable to classify whether the material has defect or not. Basically this project’s aim is to substitute commonly used Statistica software by a special application software solution developed in C#.