Application of an artificial neural network to the prediction of firedamp emissions in coal mines - Archive ouverte HAL Access content directly
Conference Papers Year : 1999

Application of an artificial neural network to the prediction of firedamp emissions in coal mines

(1) , (2)
1
2

Abstract

Coal extraction at great depths and high output faces lead to more and more elevated and irregular firedamp emissions. To respect safety conditions, coal production must be programmed to limit methane emissions in airways. In this context, the prediction of firedamp emissions is an interesting tool to optimize safety and production. But mathematical modelling of firedamp desorption and gas circulation physical processes involve non-linear physical laws and a high number of hardly accessible parameters. So artificial neural networks have been developed to model firedamp emissions : artificial neural networks as universal approximators are able to learn from examples and generalize in unknown situations. Artificial neural networks need a large amount of data sufficiently representative to learn the physical processes. Data relative to mining ventilation, such as methane concentration and air velocity in airways, are monitored and can be used to model firedamp emissions. The model based on artificial neural networks has been calibrated and validated using data from coal faces recently exploited in Lorraine Coalfield (East of France). The model reliability has been appreciated on the results of a posteriori forecasts. The model is used to forecast methane concentration values in airways as a function of coal production.
Fichier principal
Vignette du fichier
1999-072_hal.pdf (477.63 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

ineris-00972184 , version 1 (03-04-2014)

Identifiers

Cite

Jean-Christophe Couillet, Marc Prince. Application of an artificial neural network to the prediction of firedamp emissions in coal mines. APCOM Symposium 1999, Oct 1999, Golden, United States. pp.933-940. ⟨ineris-00972184⟩

Collections

INERIS
115 View
305 Download

Share

Gmail Facebook Twitter LinkedIn More