Evaluation and calibration of low-cost particulate matter sensors for respirable coal mine dust monitoring

ABSTRACTProlonged exposure to coal mine dust has led to various respiratory diseases among coal mine workers. Accurate monitoring of personal exposure concentration is crucial for evaluating dust exposure among underground coal mine workers. However, existing personal monitoring devices are expensive and bulky, limiting their widespread application and resulting in unknown personal exposure levels for most miners. This study evaluated the performance of two low-cost particulate matter (PM) sensors (PMS12 and PMS16) at high and low coal dust concentration levels. The results showed that both sensors possess good enough linearity (R2=0.92-0.93) and acceptable errors (RMSE = 29-32μg/m3) at low concentrations. However, as the coal dust concentration increased, linearity decreased (R2=0.76-0.79), and measurement errors increased (RMSE = 238-244μg/m3), but the NRMSE was still below 21%, similar to those at the low concentrations. Precise detection of experimental results showed that the sensor's precision level diminished with rising coal dust concentrations. To further improve the PM sensor's measurement performance, a two-layer correction model was introduced. It incorporated the two top-performing models chosen from the four basic models (KNN, RF, ET, XGBoost) as inputs for the second-layer linear regression model to make predictions. Additionally, temperature and humidity data were included as correction factors in the model. The results indicated that the correction models exhibited strong performance across all levels of coal dust concentration (R2 was 0.97-0.98, and RMSE was 80-91μg/m3). This study demonstrates that the application of low-cost sensors for personal coal dust exposure monitoring in underground coal mines is feasible with appropriate calibration.DisclaimerAs a service to authors and researchers we are providing this version of an accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proofs will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to these versions also. Additional informationFundingThis work was supported by National Natural Science Foundation of China (Grant No. 52074274) and the Fundamental Research Funds for the Central Universities (Grant No. 2021YCPY0107)