Stručni rad
https://doi.org/10.37023/ee.12.1-2.5
STATISTICAL/MACHINE LEARNING FOR SURROGATE PARAMETERS AIR EMISSION MONITORING FROM INSTALLATIONS
Damir Rumenjak
; Domagojeva 9, Zagreb, Hrvatska
*
* Dopisni autor.
Sažetak
Statistical/machine learning is discussed as a part of permit conditions for monitoring emissions into air using
surrogate parameters. It is unavoidable step in establishing system of monitoring by models. Requirements for such learning,
given by Directive for industrial emissions (IED) and Conclusions for best available techniques (BATC) are recognized. They
are compared with requirements in standards for direct continuous emission monitoring and automated measuring systems
and then use to broadly define statistical learning seeking the common principles that could be applied in permits. Such are
found to be clear phases of learning introducing training, validation and testing, basic equations for learning, learning paths,
blocking of observations and quality assurance based on statistical criteria. The findings are intended for monitoring practice
using continuous monitoring of air emissions (mineral and energy sector and for waste incineration and co-incineration and
even broader), needs permitting procedure and could use surrogate parameters models for monitoring emissions into air.
Ključne riječi
1st Statistical/machine learning; 2nd Emission monitoring; 3rd Surrogate parameters
Hrčak ID:
341606
URI
Datum izdavanja:
18.12.2025.
Posjeta: 198 *