Quantitative analysis of dust pollution characteristics and influencing factors in mining areas based on statistical modelling

Abstract

As disorderly dust emission of open-pit mines often leads to ecological degradation, this study therefore conducts a quantitative analysis on the evolution patterns of particle changes and the influence of environmental indicators on the weights of microdust. Taking the main regions of dust occurrence in Hequ open-pit coal mine as the subject for research, this study uses the dust monitoring system to obtain data on TSP, PM10, PM2.5 and environmental indicators in different target regions. We conducted comparative analysis on the differences in the distribution of particles with different sizes based on dust concentration and introduced the Institute for Administrative Quality Improvement method, Pearson correlation matrix analysis and Grey Relation Analysis to explore the core pollutants, the intrinsic correlation of dust with different particle sizes, and the correlation between environmental indicators and dust concentration in the target regions. Based on the univariate regression analysis, MLR and PCA-MLR, we verified the predictions of the MLR and the PCA-MLR model by MRE. The results show that: The concentration of dust with different particle sizes in Region No.1 (excavation site) and Region No.3 (coal yard) exceeded the secondary limit in the current standard, while the concentration in Region No.2 (traffic artery) only exceeded the primary limit. ②The dust pollution capacity of different regions was consistent with results from IAQI assessment: Region 1>Region 3>Region 2, where the core pollutants were all PM2.5. When the concentration of TSP was consistent in different regions, we found Region 2>Region 3>Region 1 in terms of their dust pollution capacity. ③Dust concentrations in different areas were found to be linearly significant. ④The patterns of fitting in multivariate linear equations based on MLR models of different regions tended to be consistent with the Pearson correlation of dust concentration, with the multivariate fit outperforming the univariate fit. We also found Region No.3 (3.02 %)> Region 2 (9.46 %)>Region 1 (10.75 %) in terms of the prediction accuracy of MLR model.⑤TSP and PM10 were strongly positively correlated with barometric pressure while PM2.5 was strongly negatively correlated with relative humidity in Region No.1;dust with different particle sizes was strongly negatively correlated with temperature and wind speed in Region No.2 yet only negatively correlated with temperature in Region No.3.⑥The PCA-MLR model outperformed the direct MLR model with a 56.63 % and 13.41 % increase in microdust weights and environmental indicators.

Keywords

• dust concentration
• pearson correlation method
• gra analysis method
• environmental indicators
• pca-mlr modelling