Research on wind-spray synergistic dust removal technology in coal mine

Abstract

The high concentration dust produced by coal mining seriously endangers the health of coal miners. This study presents an integrated computational-experimental framework for optimizing a wind-spray synergistic dust collector to enhance dust suppression efficiency at the coal face. An improved Euler-Lagrange model incorporating O’Rourke’s droplet collision algorithm was developed, reducing prediction error by 6.7 % compared to standard models. Systematic CFD simulations investigated the effects of seven operational parameters (spray pressure, nozzle diameter, fan velocity, and spray angles) on spray field characteristics. To enable comprehensive multi-parameter optimization, four machine learning (ML) algorithms were evaluated using a 500-sample dataset generated through Latin Hypercube Sampling. The Back-Propagation Neural Network (BPNN) demonstrated superior predictive performance (R2 = 0.981, RMSE = 0.33 g/m3, MAPE = 4.8 %) with excellent generalization capability (ΔRMSE = 0.03 g/m3). Global optimization employing the validated BPNN surrogate model with Sequential Quadratic Programming refinement identified optimal parameters, achieving 136.4 % improvement in average droplet concentration near the drum compared to baseline configurations. Field implementation of the optimized dual-collector system reduced dust concentration at the shearer driver’s position from 372 to 43.7 mg/m3, demonstrating 90.6 % removal efficiency. This integrated CFD-ML methodology provides a robust framework for industrial dust suppression optimization, with significant implications for coal mining safety.