A Novel Method for Predicting Tunnel Blasting Dust Yield Based on Numerical Inversion and WOA-DELM (Whale Optimization Algorithm-Deep Extreme Learning Machine) Neural Network

Supplementary Materials List and Descriptions1. Supplementary Figure S1
Suggested File Name: Supplementary_Figure_S1_Data_fitting_of_blasting_dust.tifDescription: This figure presents the data fitting results of blasting dust concentration for four test sections (Sections 1-4). The four subplots (a, b, c, d) display the time-series fitting curves of dust concentration under varying conditions of surrounding rock type, temperature, humidity, and water content grade. It visually demonstrates the alignment between field-measured data and the theoretical model, serving as the direct basis for the "Fitting value" data in Table 2 of the manuscript.2. Supplementary Figure S2
Suggested File Name: Supplementary_Figure_S2_Prediction_of_dust_yield_and_single_factor_variation.tifDescription: This figure illustrates the single-factor sensitivity analysis of tunnel blasting dust yield based on the WOA-DELM neural network. It clearly shows the sensitivity (change per unit) of the predicted dust yield to variations in six key input parameters: surrounding rock type, temperature, humidity, material moisture content, distance from the monitoring point to the tunnel face, and distance from the ventilation duct outlet to the monitoring point. This is the key visualization supporting the conclusions of the sensitivity analysis presented in Table 6 of the manuscript.3. Supplementary Figure S3
Suggested File Name: Supplementary_Figure_S3_Performance_Comparison_of_Predictive_Models.tifDescription: This figure, composed of multiple subplots (a-f), provides a systematic performance comparison of various predictive models (including WOA-DELM, WDO-DELM, GWO-DELM, PSO-DELM, ABC-DELM, MVO-DELM, ASO-DELM, DELM, BP, and KELM) across multiple evaluation metrics (RMSE, MAPE, MAE, R², IA, PRIA). It serves as the core comparative evidence validating the superiority of the WOA-DELM neural network model, which achieved the highest evaluation function value of 0.939.4. Supplementary Table S1
Suggested File Name: Supplementary_Table_S1_Inversion_value_of_tunnel_blasting_dust_yield.xlsxDescription: This table provides detailed data on the inversion values of tunnel blasting dust yield for the four test sections. It includes parameters for each section: surrounding rock type, excavation volume, tunnel water content grade, temperature, relative humidity, fitted blasting dust output, inverted blasting dust yield, and the inverted yield per unit volume. This dataset forms the foundational data connecting field measurements and numerical inversion to the final corrected formula (Equation 29) and supports the analysis and discussion in Table 7 of the manuscript.