Metodika pro predikci průtoku kapalných médií s využitím nástrojů umělé inteligence

Abstract

The dissertation focuses on analyzing physical events in the cooling pipe circuit of the production unit. The main goal is to propose a methodology for predicting the flow of liquid media through this circuit. The production technology is comprehensively described, followed by a proposed methodology for creating a mathematical model, which serves as the basis for a neural network. Additionally, a SMART measuring device is designed to measure various physical quantities using its integrated sensors and evaluate these quantities through the neural network. This device also includes advanced SMART functions for self-diagnosis. For the meter's correct functionality, source code was created in C++, including a clear HMI. An experiment is clearly described, which was used to evaluate the predictions made and assess the device's contribution to scientific and commercial practice. The resulting device can be fully considered a new generation SMART sensor and is compatible with the Industry 4.0 trend.