Last modified: 2015.VII.19.14:08 - MIAÚ-RSS
A MIAÚ immár 20+ éve áll a Köz szolgálatára!
MATARKA-nézet
Optical diagnostics and analysis of steady combustion based on artificial intelligence
Leading article: 2020. November (MIAU No. 267.)
(Previous article: MIAU No. 266.)
Keywords: physics, consistence, chained similarity analyses
Abstract: Background: The cooperation between two departments (the Department of
Mechatronics, Optics, and Mechanical Engineering Informatics – MOGI and the
Department of Energy Engineering - EGR) of the Budapest University of
Technology and Economics motivated this study. EGR has a few years of
background in the field of experimental thermal and optical diagnostics of steady
combustion. Since the distributed combustion offers a low pollutant emission,
therefore, investigating such a flame with another spectrometer having higher
resolution became timely. The reduced chemiluminescent of distributed
combustion emission causes a lower intensity, requiring a more sensitive
instrument.
Challenges: Regulations for combustion chambers are continuously stringent,
especially for pollutant emissions. The emission can be influenced through
passive elements like advanced nozzle geometry and/or diverging nozzle/quarl.
The effect of these passive elements can be evaluated from pollutant emission
analysis data. Besides geometry, the flow field also largely influences emissions.
The corresponding features include, e.g., flame shape, pressure, andtemperature
fields. Control of these parameters is usually realized by active systems.
However, the reaction time of pollutant emission sensors is in the range of ten
seconds, hence, an alternative solution is necessary for online control.
Tasks: Online combustion control needs real-time input about the features of
combustion, such as chemiluminescent emission and pressure fluctuations. The
former propagates with the speed of light, while the latter with the speed of
sound – both of them are excellent for the desired purpose. Therefore, the index
values were calculated firstly, based on the emitted gases derived from the
chemiluminescent signal. Both the aggregated index values about the ideality of
the emitted gases and the derivation of them based on spectral data are calculated
through artificial intelligence (AI), ensuring the expected real-time
7
characteristics. The AI is necessary to ensure a high-levelled quality management
during the automated derivation processes of production function like
Y=f(X1;...;Xn).
Results: The ideality index values of the emitted gases (as the output of AIbased term-creation processes) could be modelled based on the spectral data. The
best set of the combustion parameter-variants (the best treatment) could also be
derived based both on the dynamics of the index values and on the aggregated
views of the treatments in a validated way. The ideality index values could be
derived based on the physical aspects of the combustions like temperatures,
pressures, volumetric flow rates, etc. The correlation between the facts and
estimations of the production function (in case of the homogenous treatments):
0.83. Parallel, the different gas concentrations could also be derived, not only
their aggregated ideality values. Based on the relative robust production
function, the genetic potential of the ideality could be modelled, and so, the finetuned parameters of an ideal treatment could be derived. The AI-based models
could be produced using a high-level quality assurance system (the symmetry of
the functions), and these models have therefore a reduced gossip-potential
because these models know the none as system output too.
More (PDF)
Please, send Your comments per email!
((Back))
miau.my-x.hu
myxfree.tool
rss.services
]
