Monitoring of combustion regimes based on the visualization of the flame and machine learning

Development of modern intelligent monitoring and control systems in energy, allowing reducing the level of harmful emissions and energy intensity production is relevant. In the scientific literature usage of new efficient machine learning techniques for automatic extraction of features for the class...

Full description

Saved in:
Bibliographic Details
Published inJournal of physics. Conference series Vol. 1128; no. 1; pp. 12138 - 12143
Main Authors Tokarev, M P, Abdurakipov, S S, Gobyzov, O A, Seredkin, A V, Dulin, V M
Format Journal Article
LanguageEnglish
Published Bristol IOP Publishing 01.11.2018
Subjects
Artificial neural networks
Combustion
Energy utilization
Feature extraction
Image classification
Machine learning
Monitoring
Neural networks
Online AccessGet full text

Cover

More Information
Summary:Development of modern intelligent monitoring and control systems in energy, allowing reducing the level of harmful emissions and energy intensity production is relevant. In the scientific literature usage of new efficient machine learning techniques for automatic extraction of features for the classification of combustion regimes is insufficiently covered. In this paper we describe a method for determining combustion regimes based on images of flames. To determine the combustion regimes, a convolutional neural network is trained using labeled data. It is shown that in the gas flame colour images the accuracy of the classification of regimes is up to 98%. Results of the convolutional neural network are compared to classification results of various linear models.
ISSN:1742-6588
1742-6596
DOI:10.1088/1742-6596/1128/1/012138