Abstract
To accurately and quantitatively analyze the pollutant gas concentration in tunneling roadways, a prediction model of the pollutant gas concentration was proposed and established. Through downhole gas composition data acquisition and correlation analysis, the prediction variables of downhole gas pollution are obtained with both short-term and long-term memory neural network prediction methods and random forest regression modeling methods, making full use of historical target gas concentration data for the future in a short period of time to evaluate the model performance and prediction results. Compared with the results of the stochastic forest regression prediction and the long- and short-term memory neural network prediction, the stochastic forest regression prediction model has a good prediction effect and better generalization effect and is a reliable method with excellent performance for downhole gas concentration prediction. The analysis of the predicted results shows that the change in CO concentration is strongly correlated with CH4 and CO2 and strongly correlated with N2, making it possible to obtain the potential influencing factors of the target gas. These results provide a scientific basis for the prediction of underground pollution gas concentration and the protection and treatment of the atmospheric environment in mining areas.
Highlights
As the main energy resource in China, coal is closely related to other industries and people’s lives
The gas information in the database is put into the trained prediction model according to the above method, and the prediction is made
The long-term and short-term memory neural network and the prediction model of random forest were built to predict the concentration of harmful gases in a well
Summary
As the main energy resource in China, coal is closely related to other industries and people’s lives. Real-time monitoring of gas component changes can provide early warnings of coal mine accidents. Qiu et al. used laser spectrum technology to monitor the gas index concentration for early warnings of spontaneous coal combustion. Due to the complex downhole environment, the traditional time series prediction model has difficulty achieving ideal prediction accuracy since the prediction accuracy of the support vector machine depends on the selection of super parameters and the traditional recurrent neural network has some training problems, such as gradient disappearance. According to the traditional timing sequence prediction method and the machine learning method, the establishment of a long- and short-term memory neural network and a random forest prediction model can predict the target gas concentration and provide a scientific basis for the concentration control and quantitative analysis of underground pollution gas. The prediction ability of several models is compared and analyzed
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
