Využití moderních metod měření při kontrole metalurgických procesů

Abstract

This work discusses examples of the use of modern controllig methods at metallurgical plants and demonstrates the potential benefits of solving specific problems mainly in terms of cost savings. Work specifically deals with providing visual data of temperature measurements of hot materials and examining of these data by means of neural networks to extend base of data in production for easier solution of the root cause of defected slabs, which are intended for further processing and which have to have the desired properties. The carried out solution uses a cheap webcam that performs the video transfer into digital form, allowing decomposition of R, G, B components, which are then used as input for learning of neural networks. The paper continues by application of knowledge about the problems in designing an appropriate solution concept as a system including a computer program for determining the temperature of the material learned by the neural network. In final chapter tests of functionality are carried out of the proposed solution in real-world applications, including outlining the possible use of the solutions in education.